[session] all cnx._session become cnx.session The _session belonged to ClientConnection, which is just gone. Related to #3837233.

[repoapi] fold ClientConnection into Connection Connection replaces ClientConnection everywhere. Some notes: * testlib: .client_cnx and .repo_cnx become aliases of .cnx (we might not want to tell people to update their tests again for just no real benefit, so we'll live with these aliases for a while) * entity.as_rset must not be cached because we risk caching result sets having a Connection object as .req (helps unittest_breadcrumbs) * entity._cw_dont_cache_attributes loses its repo/request special paths and only keeps its storage/bfss user (this helps unittest_wfobjs) * moreover, entity.cw_instantiate and .cw_set stop overriding the attributes cache *after* the before_*_entity hooks have run, because there is no need to (it is now actually harmful to do it and unittest_hooks.test_html_tidy* tests remain green because of this) * rset._build_entity sticks its .req onto the entity just fetched from the cache, because otherwise it might carry a _cw that is a Connection object where a Request is expected (helps unittest_views_actions) * we get overall better cache usages (entity caches were split over Request + ClientConnection and Connection), hence the changes unittest_entity and unittest_wfobjs * void the ecache when providing the cnx to a request object Having the entity cache pre-filled when we bind it to the request object hurts because these entities are bound to Connection objects, that lack e.g. `.form` or `.add_js` and crash the views subsystem. Thus, the unittest_testlib.test_error_raised test will are kept green. Closes #3837233

[migractions] remove any session related leftovers Also, the migration handler really wants a "repo" session anyway. This highlights a bug in unittest_wfobjs.test_duplicated_transition where an old connection is reused. It probably used to work because it was made of session + set_cnxset magic dust. Related to #3933480.

[session] provide missing apis to Connection From this point, Connection provides a superset of ClientConnection. Related to #3837233.

[connection] replace .running_dbapi_query with .hooks_in_progress The thing was badly named. It tries to help distinguish between queries issued directly by the programmer (e.g in the views: cnx.execute(...)) from queries issued from the hooks, operations ... or even the repository or the native source objects. It worked heuristically being associated with the security being disabled. We provide a better name and an implementation distinct from the security management methods. Related to #3933480.

[dataimport] Turn the module into a package Just `hg mv` dataimport.py and test/unittest_dataimport.py.

[dataimport] Properly escape strings sent to COPY FROM (closes #5278743) See http://www.postgresql.org/docs/9.1/static/sql-copy.html#AEN64296 for escaping codes.

[http] fix cache-control syntax in EtagHTTPCacheManager use , to separate cache response directive closes #5307539

[connection] provide some missing documentation bits The previous changesets remove information about session data and transaction data. We provide the necessary entries in the docstring of the Connection class. Related to #3933480.

[repoapi,session] remove all session-as-cnx backward compat The `dbapi` being gone, we now can drop the session object bw-compatibility layer. This will allow further simplifications, such as folding ClientConnection and Connection (without too much pain), and then having persistent sessions. Related to #3933480.

[server] drop repo.internal_session and InternalSession

Added tag cubicweb-version-3.20.7, cubicweb-debian-version-3.20.7-1, cubicweb-centos-version-3.20.7-1 for changeset 359d68bc1260

[pkg] 3.20.7

[dbapi] retire repo.[gs]et_shared_data which were used by the dbapi Related to #3933480

[dbapi] retire repository transaction methods which were used by the dbapi Related to #3933480.

[dbapi] retire repo.call_service which was used by the dbapi Related to #3933480.

[dbapi] retire repo.commit/rollback which was used by the dbapi Related to #3933480.

[dbapi] retire repo.user_info which was used by the dbapi Related to #3933480.

[dbapi] retire repo.entity_metas which was used by the dbapi Related to #3933480.

[dbapi] retire repo.execute, which was used by the dbapi We will retire or reform pieces, mostly of the Repository class, that were related to the old dbapi way of life. Related to #3933480.

[devtools/testlib] remove dbapi backward-compat Related to #3933480.

[test] fix copy-pasta in doc string

[book] remove dbapi documentation Move some still-relevant doc related to the RQL connection API to the 'sessions' section. It might make sense to move this back outside of the "Repository development" chapter at some point since this is relevant in other contexts.

[uilib] `uilib.js` helper now honors explicit JSString (closes #4959538) Previously, if a JSString was inside a nested container (i.e. list or dict), it was handled as a string, not a verbatim js chunk (see docstring and test). So the following code:: >>> str(js.cw.pouet(1, {'callback': JSString('cw.mycallback')}) resulted in:: "cw.pouet(1, {'callback': 'cw.mycallback')})" while we expect (note the removed quotes):: "cw.pouet(1, {'callback': cw.mycallback)})" We use ``cubiweb.utils.js_dumps`` instead of ``cubicweb.utils.json_dumps`` which is aware of JSString.

[dbapi] remove the dbapi module and its immediate remaining users We suppress toolsutils.config_connect, which has currently only one known user (the email cube), which is being patched. It can be replaced with utils.admincnx function for a local access. Next will come a series to: * remove the session backward compatibility * fold ClientConnection into Connection Closes #3933480.

[web/views] remove dbapi usage from 'gc' view

[sqlite] Fix integrity error parsing Up to sqlite 3.7, integrity error message may be in the singular form if only one column is in the unique index. Closes #5224632

[postgres] fix limit_size regexp to properly remove one char tags like <p>. Closes #5221847

[postgres] _regproc.postgres.sql should be reloadable

[mail] include Date header in generated mails (closes #5271058) Mails without Date gets flagged as spam. Plus it's wrong anyway.

[doc] mention datadir-url configuration option in release notes Related to #5204550

[web] add test for datadir-url option (related to #5204550)