Added tag 3.25.4, centos/3.25.4-1, debian/3.25.4-1 for changeset b8567725c473

[pkg] Version 3.25.4

[pyramid] Let logging be configured through .ini file We do not initialize logging in CubicWebPyramidConfiguration thus preventing logging to be started with cubicweb's configuration. On the other hand, any logging configuration defined in ``development.ini`` file will be loaded by whatever start the instance using this file (i.e. pserve, gunicorn, etc.). The benefit is that logging can now be configured per "qualified name" in this file (i.e. one can easily set the DEBUG level for the application cube while keeping all dependencies' level to WARNING). In the development.ini template, we add logger configurations for "logilab" and "cubicweb" qualified names (along with those of the application cube).

Disable pytest's logs reporting CubicWeb already prints captured logs in stderr upon test failure, so we get them twice. Until we fix our logging management, disabling pytest capture and restore brings the output level back to sanity.

[pyramid] Do not issue security warnings in test mode When some session or authtk secret is missing in Pyramid settings, scary "!! SECURITY WARNING !!" are issued. This is arguably pointless in tests. So disable them in this case.

[ssplanner] Prevent execution of write queries involving computed relations Previously, setting a computed relation upon entity creation or modification (using the ORM or an RQL query) would usually fail with an operational error in the backend ("no such table"). However, under some mysterious circumstances (like passing a string as value in cw_set for a computed relation), the RQL to SQL transformation would simply drop the clause. To prevent this to happen, we add a check for computed relation before adding a relation to an execution plan. This check raises a QueryError. It happens in several places: * in querier.InsertPlan.add_relation_def() (called from several places in ssplanner steps) for INSERT queries, * in ssplanner.UpdateStep.execute() for SET queries and, * in ssplanner.SSplanner.build_delete_plan() for DELETE queries. Tests added to unittest_querier.py because unittest_sslplanner.py looked inappropriate (it has only unit tests) and the former already had a NonRegressionTC class.

[querier] Expand 'rdef' argument in InsertPlan.add_relation_def() Expand the 'rdef' tuple to improve readability. The 'rtype' value is not used at the moment, but will in the next changeset.

[server] Use dict.setdefault() instead of if/else in repo.glob_add_relations()

[cleanup] Fix some flake8 errors

[test] Stop using lgc.testlib in unittest_migractions

[test] Drop no more used "maxeid" based deletion in BaseQuerierTC and derived tests This is probably only necessary for QuerierTC itself, move it there and drop incantation from other derived classes to deactivate this feature.

[rql2sql/test] Remove dead setUpModule tearDownModule they are overwritten a few lines below , hence this code is dead and not necessary for test success, drop it.

[rql2sql/test] Move base class from repotest to unittest_rql2sql It's only used there, so move it closer to its usage. Along the way, stop using lgc.testlib.

[rql2sql/test] Set maxDiff to None to see full diffs this is usually useful to compare SQL strings

[rql2sql] Use string_type instead of unicode for py3k support

[gcdebug] Only ignore weakref.WeakKeyDictionary class on Python 2 From Python 3.5, this class fails isinstance check with the following error: :: cls = <class 'weakref.WeakKeyDictionary'>, instance = <functools._lru_list_elem object at 0x7f4331859a48> def __instancecheck__(cls, instance): """Override for isinstance(instance, cls).""" # Inline the cache checking > subclass = instance.__class__ E AssertionError: [<class 'AttributeError'> in gc] 'functools._lru_list_elem' object has no attribute '__class__' /usr/lib/python3.5/abc.py:181: AssertionError I have no clue why this happens, but it makes cubicweb.web.test.test_views.AutomaticWebTest.test_startup_views fail on Python 3.5. So only consider this class for Python 2.

[wsgi] Convert "code" passed to WSGIResponse as integer WSGIResponse is passed a Request.status_out in CubicWebWSGIApplication._render(), which, starting from Python 3.5, is an instance of HTTPStatus. However, webtest still expects a 3-digits value as status string. Fortunately, calling int() on an HTTPStatus works.

[test] Add a "version" to cubicweb_mycube/__pkginfo__.py Starting from Python 3.5, unittest_cwctl.py::CubicWebCtlTC.test_list() would fail without this.

[tox] Use py3 environment instead of py34 The former would select any Python 3 interpreter.

[querier] Allow usage of computed relation into WHERE clause of write queries This used to crash because of an attempt to access unexisting table. Closes #17113286

[cleanup] Fix some flake8 errors

[tox] Bump flake8 version to 3.5 Previous patches fixed new errors raised in this version.

[pyramid] Get rid of a couple of bare "except"

[pyramid] Get rid of a bare "except" in UpdateLoginTimeAuthenticationPolicy Also refactor a bit the method to restrict the try/except block to the code for which we actually want to catch an exception (i.e. the cnx.execute() + cnx.commit()).

[web] Avoid bare "except" in ConcatFilesHandler.concat_cached_filepath()

[test] Use ignore_errors=True in shutil.rmtree instead of try/except/pass

[server] Get rid of a bare except in sqlutils

[server] Do not use logilab.common.date.strptime in sqlutils This function is aliased to datetime.datetime.strptime() and exists only for compatibility with Python <= 2.5 which we no longer support.

[dataimport/test] Ignore flake8 error E741 in test_pgstore.py This error is about "l" being an ambiguous variable name. Here, it's arguably acceptable.

[test] Avoid ambiguous variable name "l" in unittest_req.py This is an error (E741) from flake8 3.5.

[tox] Bound flake8 version to >=3.4, <3.5 The code base is not clean with 3.5, so keep using 3.4 in *stable* branches.

[pyramid] set waitress url_scheme to https when base-url scheme is https See https://docs.pylonsproject.org/projects/waitress/en/latest/#using-behind-a-reverse-proxy Since we do not configure trusted_proxy waitress ignore X-Forwarded-Proto header and may generate http urls (pyramid request.url) when the instance is behind a https reverse proxy. This cause cubicweb-signredrequest to not work with since it rely on the url (including scheme). Set url_scheme to 'https' when CubicWeb base-url scheme is https as a workaround.

Added tag 3.25.3, centos/3.25.3-1, debian/3.25.3-1 for changeset d238badfc268

[pkg] version 3.25.3

[uicfg] Fix autoform_section rtags wrt derivation cw 3.25 introduced the 'derived rtags' feature, but missed that `autoform_section` implementation was overriding several methods where this was implemented in the base class. The following problems are tested and fixed: * during init, we should not attempt to call `_initfunc_step2` if there is some parent, * but we should handle expansion of wildcard ('*' used as subject or object of a tag) since this is handled during init step for this class, unlike others which handle this in `get` method (i.e. at lookup time), * fix overrided `get` method to consider parent rtag if any. Closes #17107020

[rtags] Enhance repr of rtags * we don't care of the class, the registry id is enough (never seen a case of sub-classing) * add the module from which the rtag is coming as well as parent rtag if any.

[web/test] Use bare unittest main instead of lgc.main

[cleanup] Fix some flake8 errors and drop upper bound of copyright along the way. autoform module is not added to flake8-ok-files since there are some remaining errors to be handled, but that's still a start.

[doc] Update the advanced tutorial to follow the new style layout of cubes among others.

[views] Respond with 400 Bad Request in case of validation error in ajax request Previously, we raised a plain RemoteCallFailed without a status code, in which case it defaults to "500 Internal Server Error". Now, for validation errors, we issue a 400 Bad Request since these are clearly client errors.

[web] fix session /cnx mix in anonymized_request

[test] Pin some test dependencies We currently have CI failures because cubes used as test dependencies have been updated to new-style cube layout. To avoid this, pin them to previous released. Those dependencies should be removed but in the mean time this should be enough (and backported in all active branches).

[test] Pin some test dependencies We currently have CI failures because cubes used as test dependencies have been updated to new-style cube layout. To avoid this, pin them to previous released. Those dependencies should be removed but in the mean time this should be enough (and backported in all active branches).

[web/form] Relax condition for non-translating field names we should test true condition instead of None, so if one raise a ValidationError using an empty string as 'no-field' marker instead of None, we don't get the PO file header that gettext likes to return when one attempt to translate an empty string.

Merge with 3.25 branch

[entities] Restore dc_ methods instead of __getattr__ proxy In 8de62610cea2, a __getattr__ proxy was introduced to handle compatibility with dc_* methods being called on entity instances instead of through the IDublinCore adapter. Unfortunately, __getattr__ does not play well with super() and since the latter is often used by, we here restore all dc_ methods with an explicit proxy to the adapter. More verbose but should work. Closes #17103999.

[pkg] add missing unittest2 dependency to spec file

Added tag 3.25.2, debian/3.25.2-1, centos/3.25.2-1 for changeset 5010381099f1

[pkg] Prepare version 3.25.2

[cubicweb/pyramid] take into account two ignored CORS configuration items from all-in-one.conf

[cwctl] upgrade: do delete the static data dir if verbosity is 0 (closes #17069749)

[doc] Fix option_*() migration commands in doc The function signature in the doc is not the same as in the code. Also option_removed should probably be removed from the doc since the function in migration.py is empty

[massive store] ensure the cwmassive_initialized table is deleted if needed This may be necessary if a previous import did badly crached, leaving the cwmassive_initialized behind.

[server] fix deserialize_schema to handle properly pg schema (closes 17076486) Postgres behaviour is to look first for a table in the configured schema then fall back to 'public' if no table is found. This must be implemented in deserialize_schema when listing existing tables to check for computed relations (cw_cwcomputedrtype) otherwise computed relations may not work when using non-default pg schema.

[server/utils] update passlib API to 1.7 (closes #17054805) while keeping bw-compat with 1.6 - 2.7 is not available for all distributions, e.g. jessie. The dependancy has been bumped in 6186cd15a46d because of a change in bc9d901cb9e6. We may easily use the appropriate method depending on the version installed.

Merge with 3.25 branch

Added tag 3.25.1, debian/3.25.1-1, centos/3.25.1-1 for changeset 5fe62978801a

[pkg] Version 3.25.1

[facets] Fix disappearance of navtop component on facet filtering which is because facets are replacing the whole #pageContent div, while this one contains other stuff than the view: * a type selector component that should be dropped for a while, * a computed title, * the page navigation. Then the view content itself is in a #contentmain div. The thing is that the navigation should be rebuilded on filtering (this is not the case for other bullets in the list above). This is currently handled specifically in the ajaxcontroller (except for the type selector which will disappear... who said it should be dropped at once?). So to fix this we: * put the page navigation into the "contentmain" div * don't replace anymore "pageContent" but "contentmain" After that we can even remove from the ajax controller the code that reimplements title handling similarly to the main template. Notice the part that changes the main template has to be ported to squareui. Closes #17074195

[web] Set response status to 400 when appropriate in ajax controller When this is clearly a client error, set status to 400. Otherwise, keep the default value for RemoteCallFailed (status=500).

[web] Only log exceptions in debug mode in Ajax controllers When we raise a RemoteCallFailed error (erroneously turned into a '500 Internal Server Error' response, but that's another business). In production environment, we do not want to log the exception in most cases where it's actually a client error. So only log the exception in debug mode.

[pyramid] Only expose 'cubicweb.bwcompat' setting for "all-in-one" configuration type The "pyramid" instance configuration does not work with "cubiwceb.bwcompat" mode (on purpose). Yet, having the setting exposed in development.ini file (generate by `cubicweb-ctl create --config pyramid <cube> <instance>` command) is misleading and we want to remove it. Thus, we only query this setting when cubicweb configuration is "all-in-one" and drop the setting line from templated development.ini file. If the option is found and True for any other configuration type, we issue a user warning (and ignore the option).

[migration] Only drop moved table entities if it does not exists For very old instances (namely docaster), this table has never been created.

[web/views] delete: show composite entities in predictible order Iterate over relations type in alphabetical order, so the order is predictible dans does not depend on PYTHONHASHSEED.

[req] fix find() generating non-rewritable rql on non final relations When filtering on a relation, find() was generating rql like 'Any X WHERE X is ETYPE, X relation EID' which work without being rewritten (it should probably not), but when applying some rewrite (eg. permissions) it raise in rqlrewrite code. def _use_orig_term(self, snippet_varname, term): ... > self.rewritten[key] = term.name E AttributeError: 'Constant' object has no attribute 'name' Generate valid rql instead 'Any X WHERE X is ETYPE, X relation Y, Y eid EID'.

[req] raise KeyError instead of AssertionError in req.find()

[req] use format() instead of % substitution For readability, avoid escaping using %% and use format() instead.

[test] make assertions on exception message

[req] clearer exception message Display what attribute trigger the NotImplementedError

[cleanup] Drop useless fake function actually not necessary.

[cleanup] Use plain unittest main in unittest_rqlrewrite

[cleanup] Fix flake8 error unittest_rqlrewrite hard one!

[rqlrewrite] Enhance detection of need for Exists node We actually want one only if this is not a (`And` / `Or`) binary tree of `Not` or `Exists` nodes, so write a recursive function to tell so. Related to #17074119

[rqlrewrite] Fix rewrite on ambiguities introduced by NOT relation or "is IN" type restriction When some inserted RQL snippet generate more solutions than the original RQL, the rewriter attempt to duplicate the snippet for each newly introduced solution. There are though some cases where we do not want this behaviour in case of ambiguities introduced by: * NOT(X relation Y) expression, since it won't be equivalent to NOT(X relation Y1, Y1 is Type1) OR NOT(X relation Y2, Y2 is Type2) ; * EXISTS(X relation Y, Y is IN (Type1, Type2) expression, since it's not actually necessary to split an explicitly introduced ambiguity (and it crash if we attempt to do so, so...). In test, we've to modify the `rewrite()` function because in the newly introduced test we need the same constraint to be applied to two variables in the original query, and this was not supported before. Notice the generated RQL in test is still *NOT CORRECT* "(EXISTS(NOT EXISTS() OR EXISTS(...))", or at least isn't optimal. This will be fixed in a forthcoming changeset. Related to #17074119

[dataimport] Add explanation about why external entities can't be inserted By default after the import processed the importer indicates which external entities can't be inserted because they are missing dependency data (other entities, used in inlined or mandatory relations). It usually helps there to find out which extids / relations are missing, so add this to the log.

[querier] Fix typo in Unauthorized message exception

[massive store] Ensure temporary metadata table get dropped even if there is some integrity errors while inserting entity values or setting back index/constraints, or any other errors. In case of exception, rollback to ensure we're not committing undesired intermediate state.

[test] Stop using lgc.testlib.unittest_main in dataimport's test_sqlgenstore.py

[test/cleanup] Cleanup flake8 errors in dataimport's test_sqlgenstore

[pkg] Fix path error in flake8-ok-files 'sources' is a subpackage of 'server'.

[cleanup] Fix undetected pep8 error

[doc] Set release date of 3.25 in changelog

[cwconfig] load appobjects subpackages recursively This is a regression appeared in 3.25.0. Appobjects (entities, views etc) subpackages should be loaded recursively, this is the historical behavior, relevant code is in logilab.common.registry._toload_info(). To keep historical behavior schema subpackages should not be loaded, relevant code is in yams.reader.get_schema_files(). We may want to have a consistent behavior later on... Closes #17073706.

[pkg] Set version to 3.26.0.dev0

Merge with 3.25 branch

Added tag 3.25.0, debian/3.25.0-1, centos/3.25.0-1 for changeset dacc5b168e29

[pkg] Version 3.25.0

[entities] Fix backward compat of IDublinCore adapter wrt dc_long_title `entity.dc_long_title()` used to fallback to `dc_title()`, and most entity types were relying on this, thus only implementing the later. Since introduction of the IDublinCore adapter, if one call `entity.dc_long_title()` on an entity that only implements `dc_title()`, it will retrieve the adapter which will then call its own `title()` method instead of the `dc_title()` method of the entity as expected. Fix this by calling instead `entity.dc_title()` which will eventually kick in the backward compat layer if necessary.

[web] Do not try to rmtree symlinks in rmtreecontent() There is a symlink in data directory (created by generate_static_dir() method) and trying to rmtree() it will fail with an OSError. So we unlink() it instead.

[pkg] Version 3.25.0rc3

[test] Stop using lgc.testlib

Fix some flake8 errors

[sources] Turn classmethod into standard method No need for a class method and easier to get proper logger including source's name.

[test] Add some tests for hooks.syncsources for parts which are not located in some source's check_config / check_urls method.

[sources] Check source's url attribute value on creation/modification Similarly as for config. Now, ldap source validation is properly done there instead of at initialization time.

[sources] Check sources configuration is fine on creation/modification Reintroduce usage of 'source.check_config' which had almost disappeared, as well as tests for the currently detected errors. Part of the system source specific checking done in syncsources reimplemented in a specific check_config implementation. Tests are dispatched among ldap / datafeed and syncsources tests but are not strictly correctly located (notably syncsources tests behaviour of the native source's check_config). The system source url checking part which disappears from syncsources will be reintroduced in a follow-up.

[test] Fix option name in ldap wrong group test which wasn't actually testing what it expected (and nothing guarantee it does now, but that's another story). Fixing this will avoid failure once we properly check the configuration.

[sources] Enhance prototype of check_conf_dict * mark it private * explicit name * update docstring Private function is still called from syncsources but this will be updated in a follow-up.

[sources] Stop translating validation error this should be done later in the web ui, where lang is properly set.

[sources] Simplify source's init method Only call it when enabled instead of giving a boolean flag indicating whether it is or not (which were not correctly considered).

[server] Deprecate Repository.sources_by_eid It's not used anymore within cubicweb itself.

[server] Add source_by_eid and source_by_uri methods to repository Most of the times we only need to retrieve one source (either by uri or eid) and querying sources_by_eid and sources_by_uri properties on repository just for one item is costly. So these methods query what's needed. We issue a ValueError (instead of KeyError for sources_by_{eid,uri} dict) in case the key is not found.

[hooks] Remove list() around repo.sources_by_uri There's no need to convert it as a list anymore since sources_by_uri is a property and will not be modified.

[server] Inline _entity_update method into init method of AbstractSource This _entity_update method does not make sense now that we do not update source from database information but always build them afresh.

[server] Drop update_config method of source It does not make sense anymore to update the config of a source instance (subclass of cubicweb.server.sources.AbstractSource) now that they are always built from database information (CWSource). In datafeed and ldapfeed, we move all code from "update_config" method in "init" method. This changeset fixes LDAPFeedUserDeletionTC.test_a_filter_inactivate() failure (unittest_ldapsource.py) introduces in previous changeset.

[server] Make "sources_by_uri" and "sources_by_eid" properties of repository I.e. do not populate these dict as repo initialization (bootstrap step) but always use information from database. This is needed because when multiple instances of the same application run, if one instance adds a CWSource the other ones will not see it. In particular, when using a scheduler instance, new CWSource will be added by the web instance and not seen by the scheduler which is supposed to update them. We thus define properties for sources_by_eid and sources_by_uri instead attributes on repository instance. CWSource entities are thus retrieved from database every time these properties are accessed. We factor out initialization of the "source" instance (subclass of cubicweb.server.source.AbstractSource) in a _sources() method. Note that this method takes care of calling "init" method on the source as well as "set_schema" (previously done in repo.set_schema(), which now only touches system_source). Accordingly the "init_sources_from_database" method is dropped along with "add_source"/"remove_source" methods. In syncsources hook, we thus drop: * SourceAddedOp operation which called repo.add_source() so that the SourceAddedHook only cares about checking source configuration now; * SourceRemovedOp and SourceRenamedOp operations for the same reason; * SourceConfigUpdatedOp as updating the live config of source is meaningless once we rely on them being retrieved from the database; * SourceHostConfigUpdatedHook hook which is now useless without call to SourceConfigUpdatedOp; In 3.10 migration script, remove usage of sources_by_uri repo attribute which, unless I'm missing something, appears useless (at least now). In tests: * unittest_datafeed: remove test_update_url method since we dropped respective hook; * unittest_ldapsource: LDAPFeedUserDeletionTC.test_a_filter_inactivate() currently fails because it still relies on live config being updated, this will be fixed in the next changeset once all "live source" logic will be removed.

Add a "Contributing" section to README with patch submission guidelines For the CubicWeb project and its dependencies, we now prefer patches submission and review by email on a public mailing list. We are thus moving away from the previous vcreview-based workflow taking place on the forge. This change is motivated by the following points: - the current reviewer assignment mechanism (pick a random reviewer, rely on reviewer availability rather than on willingness to review, send related patches to distinct people, etc.) is inefficient if not counter-productive; - most of the times, discussion only happens between the patch submitter and a reviewer with no easy way to increase the audience; - cubicweb-vcreview has no concept of patch series; - cubicweb-vcreview is not actively maintained anymore and its usability keeps deteriorating. We expect that email-based submission and review of patches will circumvent these limitations. Anybody interested in the project is welcome to subscribed to the mailing list and participate to the review process. This patch documents the basic workflow of patches submissions by email.

[cwctl] do only clean static data dir content (closes #17069762) deleting the directory itself is useless and may be a problem in automated deployment environments (user may not have permissions to delete or create that directory).

[cwctl] add a no-config-update option to the upgrade command (closes #17069607)

[cwconfig] Add "pyramid" config type to MCOMPAT dict I don't know what this is for exactly, but the accept_mode method below queries this dict with the configuration type name. And if the latter is "pyramid", it crashes since there's no key in MCOMPAT dict. So add one with the same value as "all-in-one". This code is called during "upgrade" migration in particular.

[pyramid] Avoid shutting down the repository at exit if it's already shutting down For install `cubicweb-ctl pyramid` command will explicitly call `repo.shutdown()` so the atexit hook is redundant. Prevent the latter to be executed when the repository is already shutting down (otherwise one gets an assertion error).

[server/test] Extract a "ldapsource" function in unittest_ldapsource.py We will add more use of this pattern in forthcoming changesets, so avoid repetition.

[server] Move "system" source initialization into its own method

[entities] Make CWSource's dictconfig method clearer

[entities] Move CWSource update_config method as a function in 3.15 migration script This script is the only caller of this method. In the code of the function, drop "skip_unknown" argument and continue accordingly upon OptionError.

[server/test] Turn "pull" method of LDAPFeedTestBase into a staticmethod Because it does not use reference to "self" (which should be "cls" for classmethod).

[py3] Use six.text_type instead of unicode in _CWSourceCfgMixIn.update_config()

[server] Drop condition on "3.10" migration in repo.init_sources_from_database() We'll duplicate the code of this method in forthcoming changeset so clean it up a bit of old stuff.

[pkg] Version 3.25.0rc2

[doc] More release note about session data handling mentionning deprecation of _cw.data and removal of get/set_shared_data.

[session] Drop long deprecated [get/set]_shared_data methods they are deprecated since 3.19 and access to the .data attribute itself deprecated in 3.25, so it's time to drop them.

[session] Drop the user session synchronization machinery which should not be necessary anymore since groups and properties are fetched for each request (cache lives in transaction_data).

[session] Avoid deprecation warning on access to Connection.data this attribute has been deprecated during the session related rework, remove warning by explicitly storing data in .transaction_data since session data is not anymore reachable through Connection object. This will allows simplification of session synchronization hooks in a later cset. Also since we do not rely anymore on session data here, we may at some point rewrite those properties to rely on the ORM instead of such custom caching.

[doc] Add release note about change in uicfg

merge 3.25 heads

[hg] Add generated slapd.conf to hgignore

Fix flake8 errors in unittest_cwctl

[cwctl] Drop 'cubicweb_' prefix when displaying cubes with 'cubicweb-ctl list' consistently with newcube / create commands. This is an alternate implementation to backed out changeset fe995d56c949. Closes #17054738

Fix flake8 errors in unittest_cwconfig

[config] Test loading of ccplugin modules to ensure ccplugin modules of cube as package aren't loaded twice, once as `cubicweb_<cube>.ccplugin`, once as `cubes.<cube>.ccplugin`. This test used to fail until fe995d56c949 has been backout. It requires calling `cleanup_sys_module` in `CubicWebConfigurationTC.tearDown` to remove any cube's submodule loaded during tests, so content of `sys.modules` may be tested.

[test] Extract out method-which-is-not-a-method mocking iter_entry_points to ease reuse and improve readability.

[cwconfig] Add a docstring on available_cubes method a renaming could be better at some point, but that's a start.

Backed out changeset fe995d56c949 available_cubes should not strip cubicweb_prefix but return actual package name for cube as package, since its output is also used to e.g. get ccplugin or site_cubicweb module name. Original aim of this set was to fix output of the "cubicweb-ctl list" command. This will be done by an alternate implementation in a later cset.

[hooks] Drop "logstats" hook It is now useless as its looping task would not run on a web instance because respective repository has no scheduler.

[hooks] Do not register "logstats" if repository has no scheduler

[server] Introduce an `has_scheduler` method on Repository This is to be used by client application to determine if looping tasks may be registered in the current process. By checking this, one will avoid the warning in looping_task method when the repository has no scheduler.

[server] Warn instead of failing when a looping task is registered and repo has no scheduler We should provide a way for client code to detect if they should register the looping task or not. See next patch for that.

[server] Exit quickly when a looping task is registered in maintenance mode In such cases the repository will not have a scheduler on purpose because the repository will not be kept running and will quickly shutdown after migration so that it's undesirable to have looping tasks being executed.

[config] Drop unused init_available_cubes method

[doc] Move content of former README.pyramid into relevant modules

Drop README.pyramid.rst file It has been kept since 3.24 when we merged pyramid-cubicweb and cubicweb-pyramid repositories into cubicweb's one waiting for someone to move the information in some canonical place but this never happened and this information is probably not so useful or already existing elsewhere. So just drop the file.

Make entypo link a true URL in README

Refer to cubicweb.readthedocs.io in README docs.cubicweb.org contains outdated documentation at the moment (and for quite some time); until it gets update, refer to the up-to-date documentation on readthedocs.

[server] don't catch exception from postcommit_event() if we are in test mode The historic behavior is to hide potential exception occurring in postcommit_event(). Unfortunately logging statements are hidden by default during tests and it become very hard to debug. At least raise if we are in test mode.

[pkg] Version 3.25.0rc1

[tox] Simplify command line to build the doc

[pyramid] Drop warning about pyramid_debugtoolbar not found in debug mode Some people way want to run a pyramid server in debug mode without the toolbar. This warning is annoying for them. Others should just install and *include* the toolbar by hand.

[pkg] Add development.ini.tmpl to "cubicweb.pyramid" package data Otherwise the file is not installed and `cubicweb-ctl create -c pyramid` does not work.

[skeleton] Don't use system site packages in tox environment As we changed this in cubicweb, there is a consensus that this is not the best practice so let us change it as well for skeleton.

[doc] Update "settings" pyramid documentation Document that without "bwcompat" we can use a development.ini file directly and add a dedicated section to list configuration options.

[doc] Update the "quickstart" page of pyramid documentation Again mentioning the two modes of operations and dropping references to prior implementation.

[doc] Rewrite "pyramid" index page in the book We drop references to pyramid-cubicweb and cubicweb-pyramid, remaining of the merge in 3.24. We now mention the two possible operation modes (with bwcompat being true or false).

[doc] Drop documentation for cubicweb.pyramid.tools module Which, as documented, mostly consists of internals.

[doc] Drop reference to "make_cubicweb_application" in pyramid doc This function got renamed into config_from_cwconfig and this is arguably an internal detail so should not appear in the API documentation.

[doc] Remove trailing whitespaces in doc/book/pyramid/index.rst

[tox] Install "pyramid" dependencies in "doc" environment Otherwise most of the cubicweb.pyramid API doc fails to build. For this we skip installation in the "doc" environment and source the new requirements/doc.txt which will install cubicweb[pyramid], thus pulling cubicweb plus pyramid's dependencies. Furthermore this requirements/doc.txt would be useful to build to documentation on readthedocs.

[web/views] fix bad html in deleteconf with show_composite=True On an entity with no composite an additional item "None" was displayed.

[web/views] display unique composite entities in deleteconf When the entity has two relations to the same composite, display it only once.

Really make rqlrewrite flake8 compliant

[pyramid] disable bwcompat by default in PyramidCWTest To avoid cubicweb exceptions handling.

Fix flake8 errors in rqlrewrite.py

[rqlrewrite/schema] Avoid parsing computed relations for each query The RQLRelationRewriter is instanciated for each RQL query, it should avoid parsing computed relations formula by using a cache, which seems rightly located on the instance's schema. This brings a *huge* performance boost to some pages on application with a few computed relations (x4 observed on a client app). Kudos to Adrien, David and Sylvain. Closes #17059828

[pkg] Restrict yams version to be less than 0.45.0 This is a backport of 10d3bf220be2 on 3.23 branch

[pyramid] Remove pyramid_debugtoolbar from "pyramid.includes" Most of the times, the package is not installed (not pulled by cubicweb[pyramid] nor pyramid itself) in a fresh environment so that first start of a new instance will fail.

[doc] Add some details about rtag derivation and server api cleanup in release notes

[doc] Simplify release notes section Those are all about backward incompatibilities, keep only one section about that.

[doc] Fill release notes for 3.25

[devtools] Create "newcube" in current directory by default Instead of using "cubes path", which in many cases will be something wrong (like <prefix>/share/cubicweb/cubes) now that cubes are regular Python packages.

[pyramid] Do not register repository.shutdown atexit in test mode It seems that there already is a special shutdown step for repository in test mode and pyramid's atexit handler interferes with it leading to assertion error about 'shutting_down' attribute being True in Repository.shutdown().

[cwconfig] don't rely on _CubeImporter while loading cubes _CubeImporter can load cube subpackages twice (there is a test for submodules in cubicweb.test.unittest_cubes.test_no_double_import() but not with a subpackage). Since _CubeImporter is for backward compatibility and is supposed to be deprecated as soon as possible, don't rely on it when loading cube schema or appobjects and always load the real modname (eg. cubicweb_<cube> if exists else cubes.<cube>).

Added tag 3.24.7, debian/3.24.7-1, centos/3.24.7-1 for changeset 58f1e8545a77

[pkg] Version 3.24.7

[pkg] Restrict yams version to be less than 0.45.0 Because of reader's API changes x and other subtle but hard to debug changes in appobjects registration in version 0.45.0. Closes #17065626.

[pkg] Version 3.25.0rc0 Let's make a release candidate version of CubicWeb 3.25.0 (only for the Python package), in the hope that some issues that we usually discover after release will be sorted out before.

[pyramid] Rename paste app factory to pyramid_main Making it clearer that this would build a Pyramid application, just in case we eventually provide other means to create a WSGI application than Pyramid.

[tox] Do not use site packages in testenv This is the normal way of running tox. We used to set this because some people liked to run tests using packages installed by other means (i.e. system-wide). Now, there seems to be a consensus that running tox without site packages is a better idea, in particular since many of our dependencies are now available as wheels and thus would install quickly provided one uses a recent pip/virtualenv.

[test] Add user-site to PYTHONPATH before installation of new cube in test_newcube_install Otherwise, setuptools complains that the target installation directory (i.e. "user site" is not in PYTHONPATH). This is probably because we use a fake HOME with --user install option. Failure shows up when running tests through tox without system site packages (see next patch).

[pkg] Depends on yams >= 0.45.0

[pkg] Depends on logilab-common >= 1.4.0

Merge with 3.24 branch

Added tag 3.24.6, debian/3.24.6-1, centos/3.24.6-1 for changeset 4029fb83a67d

[pkg] Version 3.24.6

Merge public heads

[web/views] deleteconf: ignore workflow when displaying composite entities This hide TrInfo composite entities when deleting an workflowable entity.

[rset] Use RQL cache in syntax_tree method This avoid several unnecessary parsings per HTTP request, since in most case rset's rql is in the rql cache (since it has just been executed), while yapps based parsing is known to be slow.

[repo] Introduce a clear_caches method on the Querier class that is called from repository's clear_caches.

[repo] Consistent API for cache clearing clear_caches, reset_caches, clear_eid_caches are now all named 'clear_caches' with optional eids/etypes arguments to clear eid specific cache entry,

[repo] Extract rql cache handling to a dedicated class

[sobjects] Fix a trivial flake8 error

[repo] Drop cache clearing methods from the source interface Those are actually system source specific.

[repo] Move and rename repo._clear_planning_cache to make it clearer it's clearing the @cache of source_defs method.

[sobjects] Fix flake8 errors in services.py

[test] Rename BaseQuerierTC._access to BaseQuerierTC.admin_access so it's consistent with CubicWebTC and avoid access to a protected attribute which has been exposed by the removal of the 'session' property a few csets earlier.

[repo] Fix flake8 error

[pyramid] Add a "pyramid" instance configuration type In a new module 'cubicweb.pyramid.config' we define a "pyramid" instance configuration type. The noticeable feature of this configuration is that it manages a 'development.ini' file that gets installed in application home (along with `.conf` file). This file is templated and includes generated values for secrets of session and authtk tokens. This means that we can just call: pserve etc/cubicweb.d/<appname>/development.ini or gunicorn --paste etc/cubicweb.d/<appname>/development.ini -b :8080 just after instance creation to get a pyramid instance running without having to hack around a 'pyramid.ini' file. This patch drops 'development.ini' from skeleton and moves it in cubicweb/pyramid so that it gets installed at instance creation which is more appropriate than in cube creation. The new configuration class sets "cubicweb.bwcompat" setting to false so it is not intended to replace the "all-in-one" configuration type (which would require a bit more work). This configuration is close to the the 'repository' configuration type with just a couple of options from WebConfiguration that are needed for Pyramid (anonymous user/password plus some miscellaneous options that I'm not so sure are really needed). Note, in particular, that we do not pull CORS settings to be injected as a WSGI middleware like in wsgi_application_from_cwconfig() since I believe this should be left as an end-user responsibility and since this can be defined in a standard way in paste configuration. This configuration inherits from ServerConfiguration but registers the same appobjects as WebConfiguration. In cubicweb.web.request._CubicWebRequestBase, we guard against access to "uiprops" and "datadir_url" of the config because this new "pyramid" config does not have these (this does not make sense without bwcompat mode). At some point, we should either avoid using `cw_request`'s pyramid request attribute or make cubicweb's web request really independant of existing implementation and drop these assumptions.

[pyramid] Call repository's shutdown method atexit This way when run through a WSGI server (pserve, gunicorn, etc.) we make sure that the repository is properly shut down. In particular, this would wait for threads to terminate (otherwise, they'll be killed).

[querier] Turn repo.querier_cache_key into a private function of the querier module, only place where it's used.

[cleanup] Fix flake8 errors in web/application.py and associated test

[test] Simplify CWTC's init_authentication and assertAuthSuccess which shouldn't rely on removed .session attribute while it's not necessary and could be written in a simpler way.

[cleanup] Drop non sense sentence in docstring

[repo] Drop Session at last This is not used at all anymore.

Fix flake8 some errors

[test] Use existing reference to repo_source

[repo] Drop repo.new_session method we should not go through Session to create Connection anymore. This is unofficial API, no backward compat for now.

Stop using Session on the repository side Only expect session on web request, and let the web session/authentication managers provide them. Access to cnx.data, which used to return session data, is deprecated: there is no more access to session data from the repository side, and they should be access from req.session.data from the web side.

[test] Drop unnecessary call to new_session This is not unittest, the above call is enough.

[user] Drop ugly hack on CWUser entity class which should be not necessary anymore, beside two remaining usage in our own test suite.

[test] Use user_session_cache_key function to generate cache key.

[session+test] Stop storing / accessing session when it's not necessary

[devtools] Stop relying on Session to create connection in RepoAccess

[devtools] Stop using repo.new_session

[session] Kill support for a 'session_open' hook event To be consistent with dropping of the 'session_close' event, and also because this is already not called when using pyramid.

[devtools] Rename FakeSession into FakeConnection because it is what it is.

[cleanup] Drop references to the old internal_session method dropped a while ago, no need to talk about it in a docstring nor to fake it.

[session] Drop unused class

[cleanup] Drop no more used parse_repo_uri function

[server] Fix flake8 error of test/data/hooks.py

[repo] Drop repo.connect backward compat use new_session instead (introduced in 3.19 iirc)

[session] Drop no more necessary Session.sessionid and Connection.connectionid

[repo] Kill repo._sessions and with it the need to have a looping task to clean it up, and with that all the session's timestamping machinery.

[repo] Stop closing session The only point in "closing" session is to call the `session_close` event. Since I'm not aware of any application relying on it, I think this is too costly (in term of code) to maintain and propose to drop it.

[views] Stop giving stats / debug information relying on _sessions This is already broken when using pyramid anyway.

[session] Stop relying on _sessions to synchronize living session by doing what we currently do in pyramid. Also, attempt to synchronize CWUser's class sounds like a dark corner case handling that should be removed in favor of short lived user objects (beside it shouldn't occur except during migration).

[test] Drop useless access to repo._sessions test still succeed afterwards.

[repo] Fix flake8 errors in server/__init__.py

[repo] Drop no more used ShuttingDown exception

[repo] Drop _get_session method it's not actually necessary but holds underlying _sessions dict handling. Drop backward compat relying on it (which IMO should not be a problem). Drop import of unused QueryError along the way.

[test] Drop call to CubicWebConfiguration.load_cwctl_plugins in unittest_cwctl.py This is no longer need now that we call load_available_configurations() in the "list" command (see 5e7282bdf140).

[web] Extract a base class out WebConfiguration This will be used to build a CubicWebPyramidConfiguration in following changeset.

[tox] Drop reference to removed init_instance.py file in flake8-ok-files.txt The file got removed in 1b9fd8bda1cd.

[web] Put 'use-uicache' option in level 2 So that users do not get prompted with this. Follow-up on c34590161082.

[server] Add a "scheduler" command to run repository scheduler This commands starts the repository scheduler as a standalone process that should complement a CubicWeb web instance running as a WSGI application. Added a log message in repository's shutdown method to help testing the command (i.e. make sure the method is called after the scheduler stopped). Related to #17057223.

[cwconfig] Do not instantiate a repo with a scheduler in repository method None of the caller of this method need the repository to have a scheduler and having a scheduler should be something that's explicitly requested.

[server] Make cubicweb/server/utils.py flake8-clean

[server] Drop utils's LoopTask and TasksManager classes not used anymore Related to 17057223.

[server] replace repository's tasks manager by a scheduler Repository() does not accept anymore a 'tasks_manager' argument but rather a 'scheduler' argument which is expected to be an instance of sched.scheduler class. The drop the _tasks_manager attribute of the repository and adjust all internal usages of it. In particular, in the 'repo_stats' service we do not export 'looping_tasks' statistics anymore as there's no way to retrieve this anymore from a web instance. Closes #17057223.

[server] introduce a scheduler class to run repository "looping tasks" We just use the sched module from the standard library and introduce a tiny Python2/3 compatibility layer (more for convenience actually). The "looping" aspect of tasks (previously in LoopTask class) is re-implemeted as a `schedule_periodic_task` function. This is a reasonably thin layer as compared to LoopTask/TasksManager classes. Only the "restart" aspect of LoopTask is no longer present as I'm not sure it's worth keeping. The advantage of using this (in addition to eventually dropping our custom code) is that this scheduler class provides a `run` method that blocks the process while running tasks in its queue. So we can rely on this to have a 'scheduler' ctl command (see forthcoming patch) that would only run "looping tasks" without having to implement the "blocking" aspect ourself. Related to #17057223.

web: add options to ignore css compilation and uicache

[skeleton] call includeme from new cube

[etwist] Do not call repository's start_looping_tasks anymore and warn about this We are about to drop this method from Repository class and replace it by a blocking alternative. This is not compatible with how things currently work in a Twisted server implementation. So do not start repository "looping tasks" in Twisted server anymore and issue a warning about this. If someone is interested in restoring the "all-in-one" behavior where the repository runs within a Twisted server, they may start by implementing repository looping tasks using a Twisted mechanism such as, e.g., http://twistedmatrix.com/documents/current/core/howto/time.html and eventually provide the repository with a compatible scheduler instance so that is can register its periodic tasks. At the moment, we lack resources to do this (and maintain the Twisted server of CubicWeb in general). Related to #17057223.

[test] Flake8-clean and use stdlib unittest in cubicweb/server/test/unittest_utils.py

[web] Don't raise ProcessFormError if there is no value specified a03376213747 introduced a slight modification in None / empty string values and with it a regression where an exception is raised on empty values, while we should simply consider no value is specified and return None.

[test] Cleanup imports in unittest_form Remove unused and reorder, + use bare unittest along the way.

Fix log level parameter not taken into account in cwsource log table An error in the jquery selector of the html widget controlling the log level was the cause of this bug. Closes #15772634 (grafted from c9129aae884b476455c8ed1ad802efc3a26503a5)

[pyramid] Drop call to repository's start_looping_tasks method and warn about this We are about to drop the start_looping_tasks method and running "looping tasks" along the WSGI application is not a very good idea. So issue a warning pointing to the forthcoming "scheduler" command. Related to #17057223.

[test] Make unittest_serverctl.py flake8-clean

[wsgi] Drop calls to repo.start_looping_tasks It's unlikely that these implementations were ever used so we do not bother with deprecation or a warning. We just drop the call to this method which should not be called from within a WSGI application. Related to #17057223.

[pyramid] Drop useless "init_instance" module

[server] Drop inexistent "bootstrap" argument in Repository initialization This is a missing bit from 5de78b6fff2e. Unfortunately this code appears to be uncovered by our test suite.

[server] fix a typo in a docstring

[ctl] Load available configurations in "cubicweb-ctl list" Otherwise we rely on associated ctl plugins to be loaded which is arguably orthogonal to configurations.

[web/views] Fix a docstring

[server] Use looping_task method in Repository._prepare_startup()

[skeleton] Drop logging configuration from development.ini This is actually not useful as logging is already initialized and configure "somewhere else" where the repository is started.

[pyramid] Drop retrieval of auth/session secret in all-in-one.conf We actually never supported this in cubicweb, but only in pyramid-cubicweb.

[cwctl] Recommend 'pyramid' before 'wsgi' when twisted is not available Not sure anyone ever really used the latter and the former is certainly more reliable/tested.

[cwctl] Drop message about 'the pyramid cube' in "start" command This is shown when twisted is not importable and we suggest to use 'cubicweb pyramid <appid>' instead. Since 3.24, cubicweb-pyramid got merge in cubicweb so drop this reference in the message.

[cwconfig] Fix two NameError in load_available_configs() This comes from a copy-paste in d92d75b17a85.

[etwist] Drop try/except around definitions of 'all-in-one' configuration and command handlers cubicweb.server is no longer conditionally shipped, neither in python package nor in debian package since 3.24.

[server] Call 'server_startup' hooks at the end of repository "bootstrap" Now that we have guarded all calls to Repository.looping_task() in 'server_startup' hooks we can execute this hooks category at bootstrap step. This way, repositories running without a tasks manager (i.e. those embedded into a WSGI application) will have these hooks triggered.

[pyramid] Drop module-level cache and cleanup looping tasks in tools And use a LRU cache over cached_build_user function. This looping task is problematic because it would not be run from within a WSGI application which does not have a repository with a tasks manager. This pulls an explicit dependency on 'repoze.lru' but it's not a big deal since pyramid already depends on this. RPM spec file not update since it does not even mention pyramid...

[hooks] Return early in server_startup hooks if the repository has not tasks manager

[server] Only register "clean_sessions" looping tasks if repository has a tasks manager This prepares for calling Repository's _prepare_startup() in bootstrap() so that repository instances without a tasks manager running can still have their 'server_startup' hooks triggered (this is particularly useful for BFSS storages).

[utils] Do not pass a tasks manager to Repository in admincnx() We do not call repo.start_looping_tasks() for the instantiated repository, so a tasks manager is useless.

[server] Separate repository bootstrap from initialization Thus we now call 'bootstrap' here and there explicitly and remove call of this method in Repository.__init__(). This way instantiation of a Repository does not *implicitly* triggers the "bootstrap" step, which is arguably not a trivial thing and thus deserves to be independent. In __init__, set 'shutting_down' attribute to None and then to False in bootstrap as a mean to indicate that, when initialized, a repository is neither shutting down nor started (not sure where this is used though).

[server] Also remove tmpdir in case of error during restore database command

[server] Rename Repository's init_cnxset_pool method bootstrap See the (removed) docstring for a rationale...

[server] Replace server config's init_cnxset_pool attribute by a "bootstrap" parameter in Repository This 'init_cnxset_pool' class attribute is actually used to control whether a Repository instance should be "bootstraped" (see Repository.init_cnxset_pool() for a definition of "bootstrap") or not. I seems clearer to me to have this controlled by a boolean "bootstrap" initialization parameter in Repository.

[server] Drop close/open steps in NativeSQLSource.restore() The only place where this is called is in ServerMigrationHelper.restore_database() where config.init_cnxset_pool is set to False. So these steps appear to be useless.

[skel] Fix rpm's .spec file skeleton It's broken for new-style packages since it doesn't consider (even remove) python packages.

[server] Move "starting repository..." message where this actually happens

[server] Initialize Repository without a tasks manager in ServerMigrationHelper.restore_database() We don't need a tasks manager here I think. Thus instantiate the Repository from its class directly instead of using config.repository() which set a tasks manager.

[server] Fix a typo in log message in Repository.init_cnxset_pool()

[cwconfig] Issue a warning if a configuration module cannot be loaded We already do this when loading ctl plugin modules, so make things symmetrical for config modules.

[ctl] Load available configurations in "cubicweb-ctl list" Otherwise we rely on associated ctl plugins to be loaded which is arguably orthogonal to configurations.

[skeleton] Set cubicweb.bwcompat to False in development.ini Quite a few things from the default web UI do not work when running the application through pserve (in particular, data files are not apparently served). So better disable the "bwcompat" mode until we are sure it works properly (or it gets dropped ;)).

[pyramid] Use existing repo in PyramidCWTest We thus do not make use of config_from_cwconfig() and build the configurator instance by hand prior to include 'cubicweb.pyramid'.

[pyramid] Make it possible to setup CubicWeb instance from an existing repo in includeme()

[pyramid] Move loading of pyramid_debugtoolbar later in includeme() To gather cwconfig and repository instantiations.

[skeleton,pyramid] Move pyramid app definition in cubicweb.pyramid module The application definition is actually not specific to the final "cube" being bootstrapped from skeleton. This patch thus move the pyramid application function into cubicweb.pyramid module and let cubicweb register the "paste.app_factory" entry point (instead of the bootstrapped cube). Useless call to `config.scan` is dropped along the way.

[pyramid] Add a copyright and docstring to all modules We add copyright statements for both UNLISH (original author) and LOGILAB.

[cwconfig] Strip "cubicweb_" prefix from cube name in available_cubes() Closes #17054738.

[web] fix os.rename usage on windows when destination exists os.rename on windows will raise OSError (or WindowsError subclass) when if destination file already exists. Also check that exception is EEXIST. There is an attempt to fix in f6ba947c but using IOError instead of OSError. Closes #14214794

[web/views] show composite entities in delete view Disabled by default, can be activated with show_composite = True Display first level of composite entities in a treeview, limited to the current (page_size - 1), so the maximum of displayed entities bump to page_size * (page_size - 1).

[workflow] Utilities for declarative definition of workflows Closes #5337897

[server] Remove pseudo-handling of exceptions in Repository._delete_cascade_multi() It seems to me that we never want to pass on an exception here be it in "test" mode or not, so remove the last `except Exception:` clause. For instance, the actual implementation would pass on an IntegrityError and thus possibly hide it in a log message. Concerning the `except Unauthorized:`, as said by the log message this should not happen because we are within a `security_enabled(write=False)` context. So remove it as well. As far as I can tell, this strange handling of exceptions dates from 37668bf302f5 and there is no clear justification to it.

[web] fix os.rename usage on windows when destination exists os.rename on windows will raise OSError (or WindowsError subclass) when if destination file already exists. Also check that exception is EEXIST. There is an attempt to fix in f6ba947c but using IOError instead of OSError. Closes #14214794

[skeleton] Add "cubicweb.session.secret" to development.ini Commented as other "secret" settings. At least, now all settings are present in the development.ini file. [ci skip]

[pyramid] Include 'cubicweb.pyramid.predicates' where it is used It's only used in "rest_api" module.

[pyramid] Rename make_cubicweb_application function as config_from_cwconfig And update its docstring. This function does not actually "make a cubicweb application", it just builds a pyramid.config.Configurator instance from a CubicWeb config object.

[server] Reprase messages of db-check to avoid using "system" and "sources" And rather mention "entity type table" and "entities" table, which are clearer references.

[predicates] Fix reference to has_related_entities in partial_has_related_entities docstring

Fix possible double import of cubes modules When cubes using the new layout are imported with 'cubicweb_<cube>' and with 'cubes.<cube>', the same module is imported twice. Handle this by adding 'cubes.<cube>' to sys.modules when importing from 'cubicweb_<cube>'. Move load_module() to a sub class _CubesLoader to share informations computed in find_modules(). Don't handle subpackages in _CubesImporter and rely on normal import machinery instead. Add a test and use unittest from cubicweb.devtools.testlib which resolve to unittest2 on PY2 with assertLogs() method.

Make load_module() return existing module if present in sys.modules Otherwise the reload() builtin will not work correctly: https://www.python.org/dev/peps/pep-0302/#specification-part-1-the-importer-protocol

[server] Ignore computed relations in "relations" integrity checker

[server] Use global registry variable of integrity checker functions Instead of querying globals() in checkintegrity.check(), we maintain a _CHECKERS dict mapping checker name to function in checkintegrity module. This is later used to build the list of available checkers in 'db-check' command help.

[doc] Add 3.25 release notes [ci skip]

[repository] possibility to disable connections pooling Add a new configuration option 'connections-pooler-enabled' (default yes). If set to no, disable connection pooling and open new connections to the database on demand. This allow to use CubicWeb with other poolers (such as pgbouncer).

[js] Fix loading image URL

[migration] Move pyramid-related migration after "entities" table modifications Otherwise, we getting into `add_entity_type('CWSession')` we try to insert a CWEType without any 'asource' column (per 3.24's schema) whereas the backend still wants it. Closes #17054035.

Merge 3.24.5 into default branch

[entities] Introduce an IDublinCore adapter Move all dc_ methods from AnyEntity to the new IDublinCore adapter and proxy them through a __getattr__ method on AnyEntity. Current test suite should be enough. Closes #3119992.

[skeleton] Add an entry point and configuration for running a Pyramid application We add a `pyramid_main` function in __init__.py that instantiate the WSGI application using "cubicweb.pyramid"; this is defined as an entry point (in setup.py) so that pserve_ can find it. Alongside comes a development.ini file which includes basic settings so that running a Pyramid+CubicWeb application works (only the "instance=<appid>" may be passed as a command-line argument). Logging is also configured there, but only includes the cube at stake and cubicweb (others could be added if needed). .. _perse: \ http://docs.pylonsproject.org/projects/pyramid/en/1.8-branch/pscripts/pserve.html

[cwconfig] Issue a warning when a "core" ctl plugin failed to load This is often because of a missing dependency, the warning would indicate this hopefully.

[dataimport] Fix "existing_relations" parameter name in ExtEntitiesImporter

[ical] Create ICal events or todos depending on component type returned by adapter

Added tag 3.24.5, debian/3.24.5-1, centos/3.24.5-1 for changeset 70d28e632206

Prepare 3.24.5

[devtools] Skip computed relations when attempting to auto-populate test database Fix the automatic database population underlying automatic test to avoid attempt to create computed relations.

[migration] Stop asking confirm to commit

[migration] Fix crash on 3.24 migration Some sql using a column droped the line above crash migration of cw < 3.24.

[devtools] Skip computed relations when attempting to auto-populate test database Fix the automatic database population underlying automatic test to avoid attempt to create computed relations.

[test] Use our BaseTestCase in unittest_rtags.py We need it DeprecatedInstanceWithoutModule for assertWarns method. Use this instead of unittest2 directly with a try/except block to get rid of flake8 error about import not being one top of file.

[server] Add a blank line before _CnxSetPool to fix PEP8 error

[pkg] Use logilab-common's master repository in dev requirements Should follow-up on 7b2247098f58 after integration on logilab-common side.

[rtags] pass module name to RegistrableInstance Instantiate a RegistrableInstance without passing the module name is deprecated in https://www.logilab.org/patch/10047069

[repository] move cnxset pool handling to a helper class The class has responsability to handle connections pool operations in a single public attribute 'cnxsets'. On Repository _get_cnxset() and _free_cnxset() are replaced by cnxsets.get() and cnxsets.release(). Drop multiple access to private attributes and methods from outside of Repository.

[pkg] Fix some error on building documentation

Merge 3.24.4 into default branch

Added tag 3.24.4, debian/3.24.4-1, centos/3.24.4-1 for changeset 35fd54c0065d

Prepare 3.24.4

[source,native] Generalize the IntegrityError constraint parsing regex This allows catching constraint-related errors that have been translated by psycopg2, replacing surrounding english quotes by french quotes + space for instance. Closes #17047951.

[pyramid] use pyramid.request `path_info` property instead of `path` path is concatenation of SCRIPT_NAME and PATH_INFO we just need PATH_INFO here

[pyramid] fix login route with language-mode = url-prefix

[devtools] Make timeout error in qunit tests configurable and use SkipTest in test_jscript.py This is grafted from 87443f279b0f in default branch (got tired of CI failing because of these unreliable tests).

[web] Handle unspecified value in TZDatetimeField `field._ensure_correctly_typed(self, form, value)` should return None if value is None. Closes #17050181.

[web] Handle unspecified 'date' or 'time' in JQueryDateTimePicker.process_field_data() dict.get() may return None, which has no attribute strip.

[cwconfig] create a virtual "cubes" package _CubesImporter now handle import of "cubes" and return a virtual module so that "cubes" is always importable without having to create a __init__.py in CUBES_DIR and add CUBES_DIR/.. to sys.path. Drop custom code in setup.py used to create CUBES_DIR/__init__.py and drop now useless warning when cubicweb is installed in develop mode Update test_cubes_path() and ensure 'cubes.file' is not loaded before testing its import. This test seems to have mistakes because cubes are loaded automatically when discovering cw commands (cubes.<x>.ccplugin), not sure how this should be fixed definitely.

Change hooks control (deny_all_hooks_but / allow_all_hooks_but) to be more predictable Prior to this, if one execute code like: with cnx.hooks.deny_all_hooks_but('metadata'): with cnx.hooks.deny_all_hooks_but(): # mycode 'metadata' hooks will be activated anyway in the inner block, which is rather unexpected (of course in real life you only see the latest hooks control statement, the former being higher in the call stack). This is due to the underlying usage of old `enable_hook_categories` / `disable_hook_categories` methods, which were introduced much before the now official context manager based API (with `cnx.[deny|all]_all_hooks_but(...)`). To move on, this patch drop the two legacy methods, rename and privatize related internal state on the connection (`hooks_mode` becomes `_hooks_mode`, `disabled_hook_cats` and `enabled_hook_cats` become `_hooks_categories`) and reimplement the `_hooks_control` context manager to simply update them. See the added unit test for details.

[cwconfig] make appobjects_cubes_modnames() public The method is used in cubicweb.cwvreg without underscore and outside of the config class, so let's make this method public. Fix autoreload with twisted.

Fix str(Unauthorized) which is really problematic on python 3 where __unicode__ is never called anymore. This was caused because in PY3, CubicWebException.__str__ was referencing __unicode__ implementation of CubicWebException, not of its subclass that implements it. Fix a flake8 style warning about lambda assignment as a bonus.

[test] avoid using a deprecated feature of yams handle_file() now require a module name (yams@d9120d7)

[test] add some tests to web.views.baseviews

[tox] Drop "touch {envdir}/share/cubicweb/cubes/__init__.py" command on py34 env This was useful when we installed cubicweb in "develop" mode but we do not do this anymore.

Drop support for https-url in all-in-one.conf This feature allowing an instance to have both anonymous and authenticated (https) url is not used anymore, let's drop it.

[test] use TemporaryDirectory context manager For consistency with others tests, use TemporaryDirectory from cubicweb.devtools.testlib.

[tox] Drop "exclude" option from flake8 config and adjust flake8-ok-files.txt accordingly Some files listed in flake8-ok-files.txt were excluded of flake8 run because of the "exclude" option in tox.ini. Some of them were non-existent files (moved), some others were actually non-flake8-compliant. In the latter case, we adjust trivial errors (blank lines, module import not on top of file) but remove others (like cubicweb/test/unittest_utils.py) from flake8-ok-files.txt.

[autoform] Go through guess_field even when field class is specified using uicfg guess_field does not only find the proper field class for some attribute/relation, but also do some basic configuration depending on e.g. constraints. Before this patch, if one was specifying explicitly the field class using uicfg.autoform_field, this automatic configuration wasn't happening. Change form.field_by_name and ff.guess_field so explicit class are also automatically configured now. Closes #14474840

[uilib] Fix incorrect serialization of python dicts into javascript objects Valid javascript object keys must be surrounded by quotes unless they are valid javascript identifiers. Valid identifiers are a defined by complex and changing specs, so it is much simpler to always use quotes. Closes #17046704.

[repository] Fix connection-pool-size not set to 1 with quick_start enabled config.load_schema() and config.init_cube() reload configuration options from config file, so a manually set connections-pool-size (eg. when quick_start is enabled) wasn't working.

[migration] Enhance assertion message to get a chance to fix the problem

[migration] Skip virtual rtypes when adding relation definition through add_cube Virtual rtypes should be skipped as they are added dynamically on loading schema. Those are skipped by e.g. schema serialisation or add_entity_type, do the same thing here.

[migration] Avoid unnecessary intermediary commit when migrating a schema Those are low-hanging fruit following changes in ac74476d686c (Fix addition of entity type including boundary constraints on its own attributes): this is no more necesary to commit the get a new relation type definition in the schema. At some point we could/should probably do more on this topic to avoid intermediary commit on e.g. entity type addition, but this requires more work.

[migration] Fix addition of entity type including boundary constraints on its own attributes This was failing because of the sequence of sql executing when adding an attribute: 1. add entity type <commit> 2. add relation type for attribute 1 <commit> 3. add attribute 1 and associated constraints <commit> etc. In the case of e.g. start/end constraint, we were trying to add the constraint before addition of the constrained attribute (e.g. add constraint on 'start' referencing 'end', but only 'start' has been added yet, not 'end'). This patch fix this by: * adding the relation type to the schema without having to commit, but keeping the operation to revert the addition if necessary - this allows to a single commit for all attributes of the entity type ; * using a LateOperation on constraint operation, so we ensure attributes are actually added before any constraint is added.

[schema sync] Refactor AfterAddCWRTypeHook and AfterAddCWComputedRTypeHook so the latter inherit from the former and enhance their docstring.

Drop embed-allowed option, gone away for a long time

Never record undo information for session or data import for session, it may causes unlimited database size expansion while it makes no sense to undo this kind of changes. For data import, they are created programatically hence we may not want to undo this either.

[cwvreg] load registry using modules names instead of directories Introspect cubicweb, cubes and apphome using pkgutil to generate the full list of modules names for loading registries. Avoiding using bogus logilab.common.modutils.modpath_from_file().

[schema] load schema from modules names instead of directories Introspect cubicweb, cubes and apphome using pkgutil to generate the full list of modules names for loading the schema. Keep historical behavior and check if source .py file exists if a module is found using python bytecode file (.pyc and .pyo) Loading schema from apphome require apphome to be present in sys.path and that "schema" module resolve to a file located in apphome. Update migraction tests to explicitely update sys.path when loading schema from different apps, use a contextmanager for this so it's more readable. Require updated logilab-common and yams

[devtools/test] Skip qunit tests in case of timeout Follow-up on 87443f279b0f where other similar tests got adjusted not to fail (in particular in our CI environment).

[autoform] Avoid two calls to field.process_form for the same field in some cases when some entity is being created and data include non-inlined relations, the values for this relation are now stored for later usage, avoiding two calls to field's process_form method, which may be unexpected for custom fields. This has been discovered in saem_ref#f5444b1f9770. Reorganize imports in the test along the way.

[pyramid] Don't use unsafe_cnx_context_manager for write queries we may have clumsy error that hide previous problems on attempting to commit the transaction. Closes #16753531

Repair database wrt indexes / unique constraints * recreate indexes dropped by 3.23 migration (but it's still unclear why) * attempt drop remaining extra indexes (there may be a bunch of these on old instances) * warn about missing expected indexes Closes #16666137

Enhance postgres index filter with values from real life This has been found on fairly old instances of ours (cwo/elo). Dunno if we really want this in, it may lead to false negatives.

Extract functions to list expected/found database indexes from check_indexes so they may be used from other contexts. Use consistent naming along the way: always use indexes, not indices.

Simplify and fix _cw.drop_entity_cache * it's never called with an eid as argument, beside in a useless case in test (removed) * the only place where it's called from outside the tests is in full-text reindexation in server.checkintegrity: we could removed the request implementation and move it in unittest_rset, byt I decided to keep it for consistency with all other entity cache handling methods * get back a fix from Julien Cristau for the connection's implementation, quoting is commit message: When removing an entity from the transaction's cache, clear the entity's own cache May avoid issues where an entity object is still accessible somewhere else (e.g. an operation) after dropping it from the transaction's cache, with a stale attribute or relation cache.

[entity/optimization] Cache rset when entity.related is called with entities=False If fail to see why we couldn't cache in this case, while this is important for optimization reason: when doing a single HTTP request, some queries are done two or three times because of predicates / uicfg or alike. Also always store tuple and not list in the cache, because: * else we get some regression * and inconsistent result type (tuple or list) * and it simply feels better to cache an unmutable object.

Ensure in 3.24 migration that entities from LDAP have consistent cwuri Closes #16666157

[doc/book] fix error in markup

[rtags] Allow to 'derive' rtags Since some releases, rtags (structure underlying uicfg) have selector and may be copied using something like: new_rtags = deepcopy(original_rtags) new_rtags.__module__ = __name__ new_rtags.__select__ = custom_selector The problem is that starting from that, both rtags wil diverge and changes in original_rtags won't be considered, while we usually want to set a few specific rules only in new_rtags. To fix this problem, this cset introduces the notion of "derivated/parent" rtag, eg: new_rtags = original_rtags.derive(__name__, custom_selector) Beside easier copying, when using the above method changes in original_rtags which are not overriden by new_rtags will be considered since it only hold its specific rules but look among its parent chain for non-found keys. Along the way, flake8 unittest_rtags. Closes #16164880

[test] Use bare unittest in unittest_rtags

[rtags] Docstring fix

[cleanup] flake8 rtags.py

[pkg] Require passlib >= 1.7.0 Since bc9d901cb9e6 we use 1.7.0 API, update packaging accordingly.

[web/http_headers] do not crash when IfModifiedSince is empty string (closes #16527954)

[pkg] Set an upper bound to passlib We're using a deprecated API and are getting the following deprecation warning: :: the method passlib.context.CryptContext.encrypt() is deprecated as of Passlib 1.7, and will be removed in Passlib 2.0, use CryptContext.hash() instead. So let's make sure we're getting a working library. [ci skip]

[cwconfig] Retrieve real path of cubes directory Since logilab-common 1.3.0, we have real path expansion modutils.modpath_from_file. But the extrapath parameter that is passed from cubicweb.schema reader may still have symlinks, so expand them here to hopefully have consistent comparison in modpath_from_file.

[server] Use CryptContext's hash method instead of deprecated encrypt method Getting rid of the following deprecation warning: :: the method passlib.context.CryptContext.encrypt() is deprecated as of Passlib 1.7, and will be removed in Passlib 2.0, use CryptContext.hash() instead.

Merge public heads

[cwctl] on upgrade, clear instance_md5_version cache The generate-static-datadir depends on a correct config.instance_md5_version(), and we just invalidated it by upgrading the instance.

[twisted] add request error handler to avoid finishing it twice And avoid stack traces like:: Traceback (most recent call last): File "/usr/lib/python2.7/threading.py", line 783, in __bootstrap self.__bootstrap_inner() File "/usr/lib/python2.7/threading.py", line 810, in __bootstrap_inner self.run() File "/usr/lib/python2.7/threading.py", line 763, in run self.__target(*self.__args, **self.__kwargs) --- <exception caught here> --- [...] File "/usr/lib/python2.7/dist-packages/twisted/python/threadpool.py", line 196, in _worker result = context.call(ctx, function, *args, **kwargs) File "/usr/lib/python2.7/dist-packages/twisted/python/context.py", line 118, in callWithContext return self.currentContext().callWithContext(ctx, func, *args, **kw) File "/usr/lib/python2.7/dist-packages/twisted/python/context.py", line 81, in callWithContext return func(*args,**kw) File "/home/me/envs/grshell-cw/cubicweb/statsd_logger.py", line 121, in __call__ return self.callable(*args, **kw) File "/home/me/envs/grshell-cw/cubicweb/etwist/server.py", line 131, in render_request code=500, twisted_request=request) File "/home/me/envs/grshell-cw/cubicweb/etwist/http.py", line 22, in __init__ self._finalize() File "/home/me/envs/grshell-cw/cubicweb/etwist/http.py", line 46, in _finalize self._twreq.finish() File "/usr/lib/python2.7/dist-packages/twisted/web/server.py", line 228, in finish return http.Request.finish(self) File "/usr/lib/python2.7/dist-packages/twisted/web/http.py", line 931, in finish "Request.finish called on a request after its connection was lost; " exceptions.RuntimeError: Request.finish called on a request after its connection was lost; use Request.notifyFinish to keep track of this.

[web/tests] Hide DeprecationWarnings

[web] remove next_tabindex() It's hard to see it work reliably in view of multiple server processes. If something like that is needed it should probably be built on the client (js) side.

[views/optimization] Ensure we call rset.possible_actions with the same argument rset.possible_actions (which should definitly not be an ResultSet method, but that's another story) has been designed to hold a cache to compute possible actions for a only once, since this may be a fairly costly operation (notably because of the 'has_editable_relations' predicates). But this has been broken by introduction of a new 'view' parameter which is not given by every call. To fix this, this cset adds the missing view argument where necessary and reimplements the rset's method to assert it's always called with the same key. Unfortunatly, those changes have to be ported to squareui and bootstrap cubes as well.

[schema/optimization] Ensure read permissions are deactivated when we're checking a permission When called from a web request, since we *are* checking some permission, read permissions don't have to be introduced in this query. We may avoid that since there now more any differences between web and repo connections, so let's do it.

[web] Enhance query log file * Add an overall page generation time * Don't log URL without any query (e.g. static files)

[massive store] Don't store eids_seq_range as a store attribute since it's not considered after object's initialization.

Flake8 bits with some copyright updates

[test] Use plain unittest in unittest_views_forms.py

Some py3k related fixes: use text_type instead of unicode

[cwctl] Kill deprecated ordered_instances method no need to keep bw compat for this.

[test/cleanup] Use plain unittest

[test/cleanup] flake8 unittest_rqlannotation.py

[debian] logilab >= 1.2.2 in requires (only in build depends) Related to #16404515.

[debian] move unittest2 to requires not recommends closes #16404515

[devtools] named context for boxes subTest for consistency

Added tag 3.24.3, debian/3.24.3-1, centos/3.24.3-1 for changeset bb5904cd284e

[pkg] 3.24.3

[cwconfig] ensure CubicWebNoAppConfiguration returns a typed value (closes #16364459)

[views/optimization] Makes has_editable_relations predicate less costly by using generator The `has_editable_relations` predicate is used to say if the 'modify' action should appear. To do so, it checks if anything is editable (ie. something in the attributes, relations or inlined section of the automatic form). This may be costly since it may have to check several permissions. To optimize this a bit, this cset turns list into generator so that we'll avoid unnecessary work as soon as we find a match.

[cwconfig] Trim down allowed exceptions in gettext language setting I can't see any valid reason to pass on ImportError and AttributeError. In particular, the latter shadowed a Python 3 error until 4f43e64603ef.

[pkg] Configure "universal" wheel build According to https://packaging.python.org/distributing/#wheels, we should create universal wheels (no 2to3 step, no C extension). Configure this in setup.cfg. [ci skip]

[rqlrewrite] Test and fix potential NameError We were referencing a loop variable, which may lead to a name error and show a potential error if there are several matching variables. To avoid this, introduce a list to hold every encountered variable and process all of them later.

flake8 uicfg

[uicfg] Remove unpredictability from autoform_section initialization It was attempting to guess smart default values by looking at the opposite relation (neg_role), but depending on the (random) ordering of the schema, it may or may not have been set yet, leading to unpredictable result with varying hash seed. Remove those fuzzy lookup all at once, it may change some (unreliable) things in your app but at least it's still a bit easier to explain. Closes #16272968

[schema/optimization] Wrap rql expressions into EXISTS node when checking individual permissions Some RQL expression may retrieve several results. It makes sense to wrap them into an EXISTS node to optimized things a bit. This is already done by security insertion of 'read' rql expressions.

[views/optimization] Prefetch state'names and comment format in wfhistory component This will avoid later queries to retrieve them.

[devtools] Make timeout error in qunit tests configurable and use SkipTest in test_jscript.py The reason is that the latter tests quite often fail in our CI environment, possibly due to concurrent execution of the browser, and this makes it hard to identify a truly failing build there. So skip tests with a timeout error.

Added tag 3.24.2, debian/3.24.2-1, centos/3.24.2-1 for changeset 296077513782 And drop erroneous 0.34.2 tag.

Added tag 0.34.2, debian/0.34.2-1, centos/0.34.2-1 for changeset 296077513782

[pkg] Version 3.24.2

Merge with 3.23 head

[migration] Fix 3.23.0 migration script, broken by a703f00718c2 * first query use "IN" where it should use "=" -> crash * second query is missing an interpolation -> crash * third query is attempting to remove constraint that have already been removed by the second query -> crash

[doc] Add 3.24 release date and reference in index pages [ci skip]

Closing "oldstable" branch We now release-based branch naming.

Closing "stable" branch We now release-based branch naming.

Merge 3.24 branch

[pkg] Bump version to 3.25.0.dev0

Added tag 3.24.1, debian/3.24.1-1, centos/3.24.1-1 for changeset da52fda6f15b

[debian] update changelog's time tag

[pyramid] Drop reference to pyramid_cubicweb in ctl command docstring [ci skip]

[debian] Rework split of cubicweb-ctl package It does not appear simple to have the .install files work. So just copy cubicweb-ctl script in DESDIR of cubicweb-ctl binary package and remove .install files. Related to #16133259.

[skeleton] Rephrase long description of Debian package into something meaningful [ci skip]

[skeleton] Depends on python-cubicweb in debian packaging [ci skip]

[debian] Have a single python-cubicweb binary for the library But keep distinct meta-packages for dependencies of the HTTP server (Twisted/Pyramid) and database backend. Related to #16133259.

[debian] Update debian packaging (closes #16133259) - use dh_python, pybuild and debhelper>=9, - refactor and simplify the debian/rules, - rename binary packages (but cubicweb-ctl) to python-xxx - remove daemon handling stuff (initscripts...) from cubicweb-ctl (one should now use a standard WSGI delivery method),

[pkg] Version 3.24.1

[tox] Use `python -m check_manifest` instead of check-manifest program To make sure we use Python from tox's virtualenv.

add debian buildpackage-generated files to hgignore

[test] Fix language negotiation test At least when executed alone (there seems to be a test isolation problem which I could not figure out).

[pyramid] No more need for an internal connection when user is cached

[web] req.lang should be None or a non-empty string

[web] Simplify a bit language handling Drop `set_default_language` and `set_user_language` by considering that: * before being bound to the user, request will have the site's default language * cnx has the user's preferred language, hence get it back to the request when it's bound to a connection The first point requires some change to cubicweb's vreg faking so it doesn't break. That should be enough.

Fix AttributeError for "lang" on repo/client connections This is a regression introduced by b48020a80dc3, which removed call to set_language from the `req._set_user` method. This is fine, but we still want a language on connections, so we've to handle it if necesary in Connection.__enter__ (i.e. once connection is properly open). When using pyramid front-end, which has a users cache, we've to cache its language as well because we must not access its preferred_language method since it's not bound to a proper connection.

Alias ugettext to gettext method of cwGNUTranslations in Python 3 The former does not exist in Python 3, but we rely on it. This fixes test_login_bad_password in cubicweb/pyramid/test/test_login.py which has never passed since being introduced in 6392f34fcdad.

[devctl] make i18n custom message extractors work with legacy layouts distname needs to be "cubicweb_<cubename>", even with legacy layouts for pkg.load_entry_point() to work closes #16272177

Fix (new) flake8 errors They showed up on upgrade of flake8/pep8.

[pyramid] Try "eid" first when retrieving a subresource of ETypeResource So that Pyramid traversal works as existing CubicWeb URL publishing.

Fix AttributeError on postcommit when db-create adds non-system CWSource entities In this case, `Repository.add_source` is called from an operation, not from `Repository.init_sources_from_database`, thus the attribute `sources_by_eid` does not exist. Closes #16251078.

[devctl] Optimize cleanup_sys_modules a bit by moving out of a loop a costly computation that is constant. This at least benefits the `i18ncube` command. Related to #15789486.

[pyramid] Install file listing monitored files in application home Instead of using a temporary file, that never gets deleted. Closes #16159807.

[pyramid] Fix login error message language Use http negotiation unless language is explicitly set for the site. Closes #16236485.

i18n update Pyramid-related messages were not translated. Closes #16236243.

[pyramid] Add action verb used in some messages displayed by the command For instance we'll see 'instance started' instead of 'instance None', which is nicer.

[pyramid] No more need to check CW version since it's now shipped with it

[pyramid] Fix 404 handling Avoid seeing a traceback in the UI by catching it before it reaches pyramid and restore usage of the '404' view. Closes #16159863

Fix broken flake8 configuration and flake8 errors which were hidden by this breakage. flake8 --filename options doesn't work as expected: * it's expected to be a shell pattern, using stdlib's fnmatch.fnmatch function internally. This funciton thinks that 'cubicweb/x.py' doesn't match 'cubicweb/x.py' (there must be a reason but that's not the point), hence no file was actually checked ; * as this is a list of pattern, each encountered file is checked against each pattern, leading to run time explosion. So maintain list of files to check in a separated file and give this list to flake8 using unix's xarg command.

[entities] Fix typo in wfobjs debug message

[doc] Add my 3.24 release notes

[migration] Drop cw_schema relation first without this, we ends up with the traceback shown at https://www.cubicweb.org/ticket/16130960. This is not the proper fix, which I have not been able to find. It seems due to this very rare case of deletion of such relation linked to CWRType vs order of execution of operation (in this case, the operation deleting the entity table is run before some other queries using it). As forcing this relation to be deleted before the entity type fixes the problem while this case seems rare enough, IMO this patch is "good enough". Closes #16130960

[hooks] Delete some properties cached on entities schema on schema updates cubicweb.schema add two additional @cachedproperty that should be cleared when in-memory schema is modified. This may leads to e.g. attempt to delete some already dropped relations. Related to #16130960

[doc] Add some words in changelog about pyramid support in core + fix its title Closes #16121152.

[pkg] Fix extra packages not installable For instance `pip install -e ".[pyramid]"` should install the pyramid dependencies. A typo caused all these extra packages to not be found by pip. Closes #16121322.

Added tag 3.24.0, centos/3.24.0-1 for changeset 9f7949b63ab2

[views] simplify and fix the TreeView (closes #16055814) Simplify the (internal) TreeView._init_params() method to get rid of the useless 'initial_thru_ajax' argument. The ajax tree view was not working properly when loaded via ajax, for instance from a lazy tab. This was due to the buggy protection against double javascript/ css inclusion, which is no more needed server-side as it is now implemented client-side. Deprecate use of the initial_thru_ajax argument of the TreeView.call() method.

[massive store] remove unused code

Drop a cw 3.24 warning

Fix unexpected req.relative_path() when calling CWTC.app_handle_request with path argument The _deprecated_path_arg decorator can not be used in this case, since it won't ensure backward compat if path value has been specified and is not similar as req's path (which is generally the case). This problem has been introduced by 904ee9cd0cf9.

[req,web/request] Move language prefix handling into web request This changes part of d8830e2bd2e0 by unsetting "lang" attribute of RequestSessionBase in particular (because it appears to break many downstream applications). The new implementation relies on a new "build_url_path" method on RequestSessionBase that is overridden in _CubicWebRequestBase to handle the language prefix in URL. As a consequence build_url in RequestSessionBase is not aware of this prefix anymore (contrary previous implementation from d8830e2bd2e0). Move respected tests from unittest_req.py into unittest_request.py. Commenting out part of test_handle_request_no_lang_negotiation_fixed_language in unittest_application.py because it now fails because of some obscure bug that cannot be resolved at the moment. Related to #15743487.

[test] Use stdlib unittest instead of lgc.testlib in unittest_request.py

[test] Distinguish "request" tests from accept parser ones in unittest_request.py

[pkg] Add setuptools to RPM build requirements

[pkg] Version 3.24.0

[req,web] Make it possible to handle page language from URL prefix Adding a short language prefix to URL (like "/en" or "/fr") changes the language the pages are displayed in. This prefix is kept during navigation. This way it is not necessary to do language negotiation, nor to use user preferences to determine which language to apply. This behavior is controlled through a new configuration option "language-mode", which replaces "language-negociation" option and which values may be "http-negotiation", "url-prefix" or "" (to disable language setting and force using "ui.language" property). Migration from previous option is not handled because I could not manage to have it working (users will get prompted with the configuration file diff anyways). Add some tests checking various scenarios. Closes #15743487

[config] fix the load_site_cubicweb() method for to 'new-style' cubes (closes #16059402) We first try to load the site_cubicweb module from the cubicweb_<cube> package, and if it fails, revert back to old cube path.

[massive store] remove on_commit / on_rollback parameters No actual use of this was found in client code. If one wants to control this behaviour, he can always override the commit method and add a try / except around flush_entities in its own store

[massive store] master_init should commit by default Default client usage is to create a master store to control slave ones. They need to access cwmassive_initialized from another transaction

[massive store] isolate SQL statements that one may want to customize i.e. provide public methods for meta data insertions

[massive store] Store entities in temporary table as well * on some entity type is encountered by a slave, create a dedicated table for insertion of entities of this type by this slave, similarly to what is done for relation - this should lower changes of conflicts in master/slaves mode ; * delay drop of constraints and indexes to `finish` method, where copy from temporary tables to regular table is done ; * insertion of metadata is done by scanning temporary tables, which may be way shorter than theier associated regular table ; * drop drop_metadata_constraints with its constraint_dropped friend attribute, there are no more necessary since this is done once in the `finish`. Related to #15538303

[massive store] Drop useless check for empty buffer If something went wrong, an error will be raised.

[massive store] Docstring / comment cleanups

[massive store] Turn a runtime error into an assertion geting there is a bug.

[massive store] Delay metadata attributes and default values handling to flush_entities This may be done once for all there, instead of for each entity in prepare_insert_entity.

[massive store] Use a slave specific table for relation insertion in the massive store * attribute a random id to the store * add a column containing this id in the cwmassive_initialized table * separate generic `master_init` which create the cwmassive_initialized table from rtype specific initialization which must be done in each slave now * delay removal of table indexes to `finish` * get back relations from each table on `finish` Similar work still has to be done for entities insertion - which will allow more cleanup to the constraints handling which is still rough for now. Related to #15538303

[massive store] docstring and __init__ cleanup

[massive store] Drop dead code This function is never used

[massive store] Reintroduce methods that are necessary to properly handle master/slave configuration Related to #15538303

[massive store] Follow configuration of the metadata generator Don't drop constraints and indexes for tables that are ignored by the metadata generator given to the store. One may now easily disable insertion of e.g. created_by / owned_by by removing them from the MetadataGenerator.META_RELATIONS set, in which case indexes for associated table won't be removed by the massive store.

[dataimport] make MetadataGenerator.META_RELATIONS customizable This should be done on the instance rather than on the class

[massive store] Rework constraint/index handling The basic idea is to keep the primary constraint on entities.eid since it's heavily used in metadata insertions. Other option would have been to drop / recreate but its very costly on big database, and the index is used for insertion into the entities table itself, so it's not worth droping it at a first glance. Also, keeping it avoids to systematically drop all constraints which depends on it. We may thus now lazily drop constraints, only on insertion of some etype/rtype for the related table. Related to #15538359

[massive store] Lazy removal of constraints and metadata indexes They should not be removed during store's init, because we may want to query the database with its index between store creation and call to prepare_insert* (e.g. to build the extid2eid map). Along the way: * rename drop_metadata_constraints into drop_metadata_indexes, because that's what it does * rework a bit impacted tests Closes #15538359

[repository] Drop the index on entities.type It should be now only used in maintenance queries in checkintegrity. Those are not worth maintaining it. Closes #15538317

[source] Drop source mapping handling It was only used by the cwxmlparser which has been deleted. This is too complex for litle benefit, we don't want to maintain that in cubicweb.

[repository] Drop the entities.extid column and associated cache This was not necessary anymore with promoted usage of the new data import API. Turn repository's _type_extid_cache to _type_cache with only the entity's type as key. This introduces an backward incompatible change: entity_metas dict doesn't contains anymore the extid key, but it doesn't seem used at all anywhere, so this sounds acceptable. Closes #15538317

[repository] Drop the entities.asource column It not used anymore thanks to previous patches. Closes #15538288

[integrity check] Stop using entities.asource column in integrity checks Simply check if entities have a cw_source, and link it to the system source when not found. Related to #15538288

[rql2sql/test] Ensure test results predictability Without this, test were randomly failing within tox due to different order of solutions.

[rql2sql] Stop generating SQL query from RQL using entities.type This was necessary in the "true multi-sources" time, it's not anymore, while maintaining this index is costly. Related to #15538317

[test] Remove some usages of lgc.testlib

[test] Avoid pytest discovery warnings Having TestServerConfiguration and alike in the test module namespace causes pytest discovery errors, because it thinks it's a test class.