evolve: move 'allprecursors()' to smarted usage The "new" smarted can be significantly more efficient than the old was. We convert our revset to take advantage of that.

evolve: move 'precursors()' to smarted usage The "new" smarted can be significantly more efficient than the old was. We convert our revset to take advantage of that.

evolve: move 'suspended()' to smarted usage The "new" smarted can be significantly more efficient than the old was. We convert our revset to take advantage of that.

evolve: move 'troubled()' to smarted usage The "new" smarted can be significantly more efficient than the old was. We convert our revset to take advantage of that.

evolve: avoid creating changectx object in _allsuccessors The repo lookups and object creations have a significant performance overhead.

evolve: avoid creating changectx object in _successors The repo lookups and object creations have a significant performance overhead.

evolve: avoid creating changectx object in _allprecursors The repo lookups and object creations have a significant performance overhead.

evolve: avoid creating changectx object in _precursors The repo lookups and object creations have a significant performance overhead.

evolve: fix the template keyworkd decorator docstring It was still talking about revsets.

evolve: search divergence within all precursors Before this patches we looking for divergence base in immediate successors only. This was a bug leading the divergence base to be reported missing more than it should.

evolve: make sure we use an unfiltered repo when looking for divergence data Let's be double sure

evolve: make sure we use an unfiltered repo when looking for divergence data Let's be double sure

directaccess: remove import from the no warning list removal of the list previously failed for some reason.

directaccess: add some commands to the directaccess list Before this patch we had a limited list of commands in the directaccess list. This patch adds about 20 commands to that list: - all the read-only commands in core supporting a rev as an argument - 'outgoing': since people can use 'outgoing' to know what is going to be pushed, the output of hg outgoing should be consistent with to the output of hg push and must therefore disallow directaccess since hg push forbids it.

evolve: move test for evolve --rev ordering in a separate file evolve --rev reordering is a complicated enough topic to justify a separate test.

evolve: properly evolve stacked unstable with --rev Before this patch, _singlesuccessor was not returning a revision number for the case where the parent of the argument was not obsolete. This resulted in bug when testing membership with the set of revisions. This patch fixes it and adds a test.

directaccess: disable directaccess for push and serve Before this patch, push and serve were subject to directaccess. This patch makes them throw error when trying to access hidden hashes.

directaccess: change rule from opt-in to opt-out Before this patch we would opt-in commands for direct access and the default filter for new repository was 'visible'. With this patch, the default filter for new repos is 'visible-directaccess-warn'. It means that by default all the commands have directaccess with warnings.

inhibit: move transaction wrapping outside of repo setup Before this patch, transaction wrapping code was done in reposetup. It happened to cause stackoverflows in repos with a lot of subreps. This patch moves the wrapping to extsetup and avoids this problem.

tests: ignores other core output in capability testing This will avoid breaking the test anytime core changes.

evolve: make uncommit respect allowunsable Before this patch, the uncommit command was performing the same way regardless of the value of experimental.evolution. With this patch if the configuration does not allow unstability, uncommit won't create instability.

directaccess: use cached filteredrevs Before this patch we were calling directly repoview.computehidden(repo) to compute the revisions visible with direct access, without going through the caching mechanism for the filtered revisions. There was two issues with that: (1) Performance: We were not leverating the cached values of the 'visible' revs (2) Stability: If there were to be a cache inconsistency with the computation of 'visible' we would crash in the branchmap consistency check partial.validfor. Consider the scenario of rebase with bookmarks: - when we delete a bookmark on an obsolete changeset (like what rebase does when moving the bookmark after rebasing the changesets) - then this changes the value returned by repoview.computehidden(repo) as bookmarks are used as dynamic blockers in repoview.computehidden(repo) - as of now, we don't invalidate the cache in the case of bookmark change - if we have a cached value from before the bookmark change, repoview.filterrevs(repo, 'visible') considers the cached value correct and returns something different than repoview.computehidden(repo) - in turn, if we use repoview.computehidden(repo) in directaccess, the subset relationship is broken and the cache consistency assertion (parial.validfor) fails if branchmap.updatecache is called in this time window This patch leverages the caching infrastructure in place to speed up the computation of the filteredrevs for visible-directaccess-nowarn and visible-directaccess-warn. Incidentally it prevents the bug discussed in (2) from crashing when running a rebase with a bookmark. Note that there still needs to be a fix in core for the case discussed in (2). The test for this side of the fix (not core's fix for (2) is very hard to implement without introducing a lot of dependencies and does not belong here. It is much easier to have the test of the fix for the scenario (2) in core along with the fix.

inhibit: improve performance of transaction wrapping Before this patch, transaction wrapping was the most expensive part of inhibit computation wise. This patch changes the revset that we use in the transaction wrapping to make it ~50x faster to compute: revset #0: obsolete() - hidden() 0) wall 0.000214 comb 0.000000 user 0.000000 sys 0.000000 (best of 11209) vs revset #0: (not hidden()) and obsolete() 0) wall 0.010965 comb 0.010000 user 0.010000 sys 0.000000 (best of 237)

evolve: extract the code computing dependencies in a separate function The code to compute dependencies between unstable changeset can be reused to compute the next reveset. This patch extracts it in its own function to make it reusable.

directaccess: add mechanism to load directaccess after some other extensions directaccess needs to load after some extensions to avoid interfering with them. This patch adds a mechanism to specify what extension directaccess needs to load after.

evolve: add another test for evolve --rev This patch adds one test for three conditions not covered before: - evolve --rev on a stack with obsolete commits without successors - evolve --rev on two sets of stacks in parallel - evolve --rev on a set of stack partially solvable (because all of the commits of the stack are not specified with --rev)

evolve: add a more complex test for evolve --rev The previous tests of evolve --rev would have passed with an ordering of revs by increasing revision numbers. This patch adds a test that would fail if that was what we were doing and therefore tests the implementation better.

evolve: add ordering of the revisions for evolve --rev When running evolve --rev we want to process the revisions in an optimal fashion to solve the maximum amount of trouble in the minimum number of steps. This patch adds a step to evolve --rev to order the revision before solving the troubles. A simple test is added to cover a basic case.

test: adapt to change in mercurial core Bundle2 is one by default and more capability have been added.

evolve: add directaccess to the setup.py directaccess was missing from the list so that when we were building evolve it was not contained in the build.

evolve: don't crash on singled out revisions Before this patch, we were crashing on unsolvable revisions when using evolve --rev. This patch removes the crash and prints a warning when we encounter such revision.

test: adapt test to less frequent branch warning The 701df761aa94 changeset in mercurial core made the branch warning issue only for the first branch created. We adapt the tests to reflect this change.

tests: adapt summary output to 6084926366b9 The 'summary' commands recently gained phases related output, 6084926366b9 put it into a "final" state. We adapt the test to this new style.

evolve: small refactoring of the uncommit function We introduce a new variable to simplify the implementation of hg uncommit --interactive.

evolve: improve error message We add a new line and change the wording from 'revsets' to 'revisions'.

evolve: move return statement at the right level When no trouble was found with the --rev option, the code of the evolve function would continue running and would solve unwanted troubles. This patch makes sure it does not happen.

evolve: remove extra whitespace This patch removes a whitespace at the end of a line.

directaccess: don't crash when evolve is not loaded Before this patch, when evolve was not loaded direct access was crashing. This patch catches the error and prevents the crash.

inhbit: don't crash on commit with no changes Before this patch inhibit would crash when running hg amend with no changes. This patch fixes this case and adds a test to prevent regression.

evolve: add a test for prune --keep -r . with active bookmark We recently changed the behavior of prune to make sure that the active bookmark stays active after prune -r . This patch adds a test for the case of prune --keep -r . that does not take the same code path.

Merge with stable

inhibit: add test to ensure that --hidden is working with inhibit There was a concern with the transaction hook in inhibit and its potentially devastating effect when used with --hidden. The transaction hook in inhibit hides all the visible & obsolete commits when the transaction ends using the visible() and the obsolete() revset. --hidden does not change the visible() revset so it is fine. If this changes in the future and --hidden actually ends up impacting the visible() revset, this test will catch it and prevent inhibit from putting inhibition markers on all of the obsolete changesets.

prune: with active bookmark should stay active hg prune with an active bookmark was deactivating the bookmark after the prune. This patch makes sure that if a bookmark was active before a "prune" operation (except prune -B) then the bookmark stays activate afterwards.

merge with stable

evolve: migrate off of now-dead util.any

inhibit: create direct access extension Since we want to use direct access without necessarily using inhibit, this patch separates both extensions. Inhibit depends on direct access and we add a test to check that it complains if that is not the case.

inhibit: handle inhibit marker on stripped revision If a revision disappear from the repo, we should not crash.

update README

evolve: don't use python sets on top of revset for evolve --rev In --rev we were wrapping the revsets in python sets. This is wrong as the use python sets force the wrapper revision set to be computed earlier than it has to and have no order guarantee. Removing this wrapping changes a test because the ordering does not appear to be the same but the end result is the same.

evolve: current bookmark wrongly moving during prune (issue4559) Before this patch: prune -r <otherbookmark> also moved the current bookmark eventhough it was not related to the pruned changes. This patch fixes it and adds a test to catch regressions.

inhibit: direct access with and without warning on a per command basis We introduce a new filtername visibile-directaccess-nowarn to enable direct access with no warning on a per command basis. The motivation behing this change is to display warning when attempting direct access in destructive commands.

inhbit: make tests work with evolution.createmarkers config Before this patch, the test for inhibit was assuming evolve fully running. For most of the features that inhbit enables, createmarkers is sufficient. This patch makes the test of inhibit run with the createmarkers config and only lifts it for operations in the middle of the stack.

inhibit: config to enable only direct access Inhibit provides several features to hide instability and direct access. We want to use direct access in evolve without hiding the instability. This patch adds a config flag to enable only direct access in inhibit, when this config flag is not set or set to false we enable all the features of inhibit that don't concern direct access.

merge with stable

bookmarks: update to use new bookmarks api via compatibility layer Tested against 3.4 and latest hg + upcoming bookmarks patches

simple4server: remove buggy wrapping of pull related function Not only the wrapping was all buggy, but simple4server is not supposed to affect client side pull.

simple4server: update bugtracker link We moved all evolution related bug to the core Mercurial tracker.

evolve: rename handlenotrouble to _handlenotrouble and add doc As suggested by Pierre-Yves on his comments on f113636997660b807c36cf1d0a06b76d53ada1a4 we make it clear that this function is private and explain what it does.

evolve: improve the help message of --rev User's are not exposed to the notion of revset, this patch makes the help message more intelligible.

evolve: add --rev option to the evolve command This patch is part of a series of patches to refactor the evolve method and make it more readable. We introduce a new --rev flag to specify a revset where to solve the troubles instead of looking from the parent changeset. All add a test checks that --rev works with several commits

evolve: extract cleanup logic in the evolve function We are going to need to reuse the cleanup logic when introducing --rev, so we extract it in a method to avoid code duplication.

evolve: simplify the evolve function We compute the troubles once and use it in two places instead of recomputing them just to count them.

evolve: optimize the computation of the troubled() revset We take the implementation from _counttroubles() and rely on the more efficient & operation for revset instead of laying down the entire set with %ld.

inhibit: don't inhibit pinned commits during rebase During a rebase we pin certain commits to always be visible. This caused the inhibit extension to inhibit them when a transaction closed. Let's make inhibit aware of such pins and not obsolete them.

evolve: extract the logic to solve one change into a method The goal is to later reuse this method to implement the --rev flag for evolve that solves the troubles in a revset.

evolve: add new variable and comment Add comment and introduce a new variable to keep track of if the progress ui is shown. Later on we will use this variable in the implementation of the --rev option.

evolve: move code to track progress and start node This makes it easier to refactor the evolve function in this serie of patches.

evolve: renaming of a variable tro was not a really descriptive name, replacing it with nexttrouble.

evolve: refactoring of the code displaying error when no trouble found Before this patch, we had many return statements in the evolve function especially in the part in charge of displaying errors when no troubles were found. We move this code in a separate function.

evolve: refactor code computing stats of obsolescence markers Duplicated code removal by extracting a function.

evolve: refactoring of code related to pushing obs markers This removes duplicated code for the pushobsmarkers logic

merge with stable bunch of cleanup on stable

update readme

evolve: update cmddebugrecordpruneparents to the new markers creation API The parents information are now first class citizen. (caught by laurent using pylint)

evolve: drop useless return This return statement is never reached. This a actually a good new because its variable was undefined. We remove it. (caught by laurent using pylint)

evolve: add missing import (caught by pylint)

evolve: improve transaction related code to use the lock module Refactored some of the places where we are using locking in evolve to use the lock module and write less code. This should also fix a couple of bug where tr could be None at release time.

evolve: remove unused variables As suggested by pyflakes, remove 5 unused variables in the evolve module.

evolve: remove unused code in _evolveany remove some code that was doing nothing useful in _evolveany.

evove: remove unused imports Three imports where unused in the evolve module.

tests: ensure we keep capturing output In 3.4, output capture with bundle2 is disabled by default on local push.

inhibit: updating to a obsolete commit prints warning message We move the inhibition marker in the same wlock than the update operation, this clears the warning message. To make it work we display the warning message on lock release in evolve.

evolve: stop conditional wrapping of _checkinvalidmarkers We support down to Mercurial 3.3, all such version have this function.

evolve: stop conditional wrapping of _getbundleobsmarkerpart We support down to Mercurial 3.3, all such version have this function.

evolve: stop conditional wrapping of _pushdiscoveryobsmarkers We support down to Mercurial 3.3, all such version have this function.

merge with stable

test-inhibit: test bookmark -D without prune enabled The --config option were dropped for lack of documentation.

inhibit: fix devel warning _inhibitmarkers was taking a transaction without the proper lock. Adding the lock removes the warnings.

evolve: add -i flag to amend command The -i flag was missing because it is not a global flag of the commit command. We can't make it a global flag yet as it is used in many places where it would not be appropriate so we just add it for this occurrence. Also, we only want to add it when commit supports it (otherwise it wouldn't work with older versions of mercurial)

prune: prune fix transaction pattern I forgot to amend.

prune: fix pruning w/ bookmarks and inhibit Previously prune would create the obsolesence markers, then try to fix up the bookmarks. With inhibit enabled, the act of closing a transaction for the bookmark step caused it to unhide commits that were trying to be pruned. The fix is to put the whole thing in a transaction.

inhibit: remove book -D dependency on cmdprune Bookmark -D was relying on the prune command to be registered, which isn't the case if we've disabled evolve commands. Instead, let's just hard code the arguments. The real fix would be to abstract away the 'I want to prune commits' functionality in evolve to a reusable function, but this will do for now.

inhibit: wrap repo.commit to inhibit nodes Rebase was failing if you rebased a commit, stripped it, then rebased it again. This is because the node was still hidden and we needed repo.commit to uninhibit it.

merge with stable

tests: ensure we keep capturing output In 3.4, output capture with bundle2 is disabled by default on local push.

update readme Mentions greg ward work in documentation.

docs: format concepts guide better, with literal blocks

docs: fix slightly unconventional .rst markup

docs: remove unused/obsolete figures

docs: remove obsolete documentation evolve-collaboration.rst replaced by sharing.rst instability.rst introduction to the concept for Mercurial developers -- no longer needed since most are now convinced obs-concept.rst largely replaced by the user guide obs-implementation.rst more for Mercurial developers; the source code is probably more accurate and up-to-date qsync.rst the extension this referred to no longer exists This fixes several more sphinx warnings. I've left some other old docs around, because they still have useful content that needs to be adapted and updated.

docs: add a table of contents This silences many sphinx warnings about documents not included in any toctree. Still need to delete obsolete .rst files to silence the rest of the warnings.

merge with stable

Added tag 5.1.4 for changeset 891c3ce894fa

prepare release 5.1.4

evolve: drop unused 'urllib' import

debugobsrelsethashtree: get the command to work again The command was untested and became broken when we introduced the new discovery hashing.

getbundleobsmarker: prevent crash from filtered content (issue4616) The discovery may discover common markers on non-served nodes. This lead to crash when building the obsmarker part.

test-wireproto: add some obsmarker discovery usage. The old code was never checking that the discovery were skipping common markers. This is now done.

test-wireprotocol: test with both bundle1 and bundle2 Bundle two is becoming the major way of exchanging things now. We duplicate the test and ensure the default version runs using bundle2.

Update debian packaging for 5.1.3

Update debian packaging for 5.0.0

create 3.1 compatibility branch The next changeset drop 3.1 compatibility.

merge with stable

Add "documentation update" to the changelog.

docs: cover 'touch' command

docs: explain that obsolete is not hidden

docs: explain --hidden option Also move the discussion of revision numbers up, and splice the explanation of --hidden into it.

docs: tweak wording, punctuation for better readability

docs: flesh out unfinished bits in the intro page

docs: revive the explanation of divergent changesets in the sharing guide

docs (sharing): more scattered wording tweaks

docs: sharing guide: clarify end-game of "Example 2" - push to dev-repo, rather than pulling into it (consistent with how we work the multiple developer scenario later) - explicitly demonstrate that public changesets are immutable

docs: minor tweaks to sharing guide (suggested by timeless)

docs: add code review scenario to sharing guide The idea is to demonstrate a simpler multiple-developer situation that does not involve getting into trouble. The final scenario illustrates Alice and Bob getting into trouble with bumped and divergent changesets by amending each other's history. The required tests and text are all written, but will need to be heavily revised because of the inserted scenario.

docs: add a bit more hierarchy to the sharing guide

docs: update sharing guide based (mostly) on marmoute's review - don't claim certain scenarios are more/less common, just simple/advanced - mention code review as a multiple-developer scenario (not described in detail yet!) - suggest "hg config --edit --local" instead of "cat >> .hg/hgrc" - use -q less often (and show resulting output) - edit some section headers to be consistent with user guide (example numbers; "amend" instead of "amending") (These are just the small changes; big changes are yet to come.)

Added tag 5.1.3 for changeset c52c15100fb2

prepare version 5.1.3

graft: avoid potential deadlock We must acquire 'wlock' before 'lock'.

commit: avoid potential deadlock We must acquire 'wlock' before 'lock'.

update readme

pull: fix bundle2 part generation The new part generator (allowing partial pull) was unproperly installed.

simple4server: cleanup debug output The debug output got duplicated with the original code. And the cleanup forgot about simple4serve. (code duplicated yeah!)

merge with stable

update readme

discovery: stop being stupid about variable name Overwritten variable are overwritten... The test coverage makes a simple pass in the loop :(

merge with stable

inhibit: explicitly test for None It is always better to check for None explicitly.

#iforgottoamendagain

discovery: port the new hashing version to simple4server

discovery: introduce a obshash1 function using fm1 during hashing (issue4587) The 'fm0' serialisation translate float to text. This give different result between python2.6 and python2.7 (because python2.7 got smarter). The 'fm1' is all binary and expected to be stable so we introduce the ability to do the discovery using hash computed with 'fm1' serialisation. We use a new wireprotocol version (Partly because I'm too lazy to check how to add an argument to the existing one) and capabilities). Support in simple4server is will come in the next changeset.

tests: use 'wget' instead of 'curl' for Windows compatibility The whole test suite now runs without skips, and the only two failures are a bookmark cache issue in test-prune.t, and a local output vs remote output reordering when pushing to ssh. The latter anyway is a core problem.

setup: allow including inhibit in the build Adds the INCLUDE_INHIBIT environment variable to allow including the inhibit extension in the build.

inhibit: Add -D option to the bookmark command The -D option for bookmark is similar to the -B option for strip. It deletes the bookmark and prunes the changes underneath it that are not reachable.

inhibit: don't leave any obsolete commit visible after closing transaction We add a callback to wrap the transaction closure to identify commits that are obsolete and visible. We inhibit them to prevent the user to see instability in the UI.

evolve: apply prune update before creating obsolescence markers Prune used to compute the destination commit to update to after creating the obsolescence markers. This is a commit to prevent a bug that would occur with the changes in inhibit for transactions (next commit). Indeed, with that change in inhibit, when creating the obsolescence markers in prune, they would get inhibited right away and we wouldn't update to the first non obsolete parent but stay on the same node as its obsolescence would have been inhibited (bug).

test-inhibit: check graph after prune We want to ensure prune keep working even with inhibit.

test-inhibit: drop trailing white space

tests: make tests work on Mac OS wc -m adds spacing before the results, we trim them to make the test pass on Mac OS.

merge with stable

inhibit: ensure that no obsolete changesets are visible after a bookmark change When creating a bookmark, we could reveal obsolete changesets. We are inhibiting the obsolete changesets resulting from a bookmark change. There are two ways to record bookmark changes to disk and we wrap both.

inhibit: ensure no visible changesets are obsolete after an update When updating to a commit we want to inhibit any obsolete commit affecting the changeset we are updating to. In other words we don't want any visible commit to be obsolete.

inhibit: enable direct access from parsing the revset tree To enable direct access we: - detect explicit hexadical reference to node in the revset tree - add these hashes to the static blockers to make them and their parent visible for the command to run

Added tag 5.1.2 for changeset e886bc501796

preparing version 5.1.2

update 5.1.2 with latest bugfix

evolve: have a stable manifest content when solving bumped We cannot really use iteritems here.

evolve: change config option name I forgot to amend...

config: allow disabling commands Now that upstream Mercurial has multiple options for obsolete configuration, we can allow enabling and disabling of evolve commands.

inhibit: Add some inhibition clearing mechanism We do not want to keep inhibition marker around for ever. So we are removing the one applying on public changeset.

add a 'inhibit' extension This extension will be used to reduce the scope of evolve in order to make some of its features available to the mass earlier. See extension documentation for details.

tests: use 'wget' instead of 'curl' for Windows compatibility The whole test suite now runs without skips, and the only two failures are a bookmark cache issue in test-prune.t, and a local output vs remote output reordering when pushing to ssh. The latter anyway is a core problem.

tests: import killdaemons.py from Mercurial 1cfded2fa1a9 This contains updated functionality to kill the server process on Windows, which is needed to run the test-*server.t tests. This version requires the pids to be passed as an argument instead of via $DAEMON_PIDS.

prune: add --keep to not touch the working copy In order to more closely emulate strip, lets add --keep to leave the working copy untouched when we do our prune.

strip: add the option for wrapping the strip command Adds an experimental option for wrapping the existing strip command and replacing its functionality with prune. It currently doesn't handle the --keep case, but an upcoming patch will address that.

wrapcommand: fix wrapping extensions that aren't enabled The extension command wrapping would previously fail if the extension wasn't enabled. Let's just eat that silently.

wrapcommand: allow adding additional arguments to commands This allows eh.wrapcommand to specify additional parameters to be added to the commands being wrapped. This will be useful when wrapping strip in the future.

pushexperiment: remove use of obsolete._enabled pushexperiment.py was using obsolete._enabled but it didn't cause any tests to fail so I didn't see it. Let's go ahead and replace it with the new obsolete.isenabled() api.

oldobsolete: remove use of obsolete._enabled obsolete._enabled is deprecated. Switch to the newer obsolete.isenabled() logic.

server4simple: remove use of obsolete._enabled obsolete._enabled is deprecated, so let's replace it's uses.

config: remove uses of obsolete._enabled Upstream Mercurial will be removing obsolete._enabled soon, so switch all our uses of it to use the new obsolete.isenabled(repo, option) function.

config: enable all obsolete options Upstream Mercurial has moved away from obsolete._enabled as the obsolete flag to a series of options. This patch makes evolve enable all the options automatically if the user hasn't enabled any already. This preserves the existing evolve experience of not having to configure anything.

merge with stable

evolve: fix manifest usage to match upstream The manifest class upstream is no longer a dictionary, so switch to using an API it does support.

drophack: fix to match upstream repair.strip repair.strip upstream now assumes the input nodelist is a list, so update to match it. The upstream change only affects the name of the backup bundle, so this shouldn't affect any other functionality.

evolve: prevent a crash in httpclient_pushobsmarkers() when pushing I've been running into a crash when pushing from my hg repo in a Fedora 16 VM to Win7 running 'hg serve', even with extensions disabled on both sides: ../hg push -r . pc pushing to http://192.168.1.4:8000/ searching for changes no changes found pushing 2 obsolescence markers (263 bytes) ** unknown exception encountered, please report by visiting ... File "hg-evolve/hgext/evolve.py", line 2482, in _pushobsolete remote.evoext_pushobsmarkers_0(obsdata) File "hg-evolve/hgext/evolve.py", line 2522, in httpclient_pushobsmarkers ret, output = self._call('evoext_pushobsmarkers_0', data=obsfile) ValueError: too many values to unpack I'm not sure how this repo differs from the one in the test suite, so I'm not sure how to craft a test for this. The failure occurs even when there _are_ csets to push. There was no crash if no obsolete markers needed to be pushed. At any rate, this code was stolen from httppeer._callpush(), where it calls self._call(). The socket exception handling wasn't necessary to fix the crash, but the calling code might as well be duplicated in its entirety. A successful push with this patch looks like this. Note the final line is _not_ in the output of the http push in test-simple4server.t: ../hg push -r . pc pushing to http://192.168.1.4:8000/ searching for changes remote has heads on branch 'default' that are not known locally: 3af110194a0c 56000e3ae44d 57ac6e51d290 7da4355c21b8 and 8 others remote: adding changesets remote: adding manifests remote: adding file changes remote: added 1 changesets with 0 changes to 1 files (+1 heads) pushing 4 obsolescence markers (525 bytes) remote: 2 obsolescence markers added

merge stable into default

Added tag 5.1.1 for changeset 9b8628e0a142

prepare release 5.1.1

update readme

evolve: use 3.2+ fullreposet to evaluate argument set (issue4515) getset() has no compatibility wrapper for old-style argument and we don't want to add more hacks in Mercurial core.

evolve: cache marker encoding to avoid performance degradation The main overhead in bug 4518 is caused by calculating relevant markers and encoding markers into the correct format. Calculating relevant markers takes (worst case) ~ 0.12 seconds. Encoding markers takes (worst case) ~ 0.06 seconds. Caching encoded markers takes care of the second part and speeds up cloning in bug 4518 by about 35%.

evolve: handle invalid obsmarkers in the `debugobsconvert` User can now recover from such situation, The error message is also changed to point the `debugobsconvert` commands.

fix release date in the readme

merge stable in default

Added tag 5.1.0 for changeset bfe9be352453

update readme and bump version to 5.1

merge default into table 3.3 is about to be released.

discovery: filter null from the discovery revset If there is nothing in common, common will be "null" and the revset will now returns "-1" crashing the discovery process. We filter out the null revision to prevent that.

tests: update to new core output Multiple output change need to be applied to the test suite.

evolve: explicitly disable bookmark on evolve (issue4432) We no longer leave the current bookmark active after an evolve, since it is no longer pointing to the working directory.

discovery: do not build sample if undecided is smaller than sample size This prevents crash with coming change in the mercurial discovery.

merge with stable

init: don't abort Mercurial on version mismatch Evolve causes Mercurial to exit if there is a version mismatch, even if no evolve command is invoked. This is a headache for people who have to work with multiple versions of Mercurial. Instead, raise an ImportError, which gives a warning but otherwise allows Mercurial to run non-evolve commands.

Added tag 5.0.2 for changeset dbd0733e5840

prepare release 5.0.2

test: adapt to core change in graft and rebase The ouput changed we have to adapt

merge with stable

doc: drop rebase extension activation The dependency have been dropped.

evolve: remove the dependency to the rebase extension We do not use any rebase internal anymore.

evolve: remove the use of `rebase.concludenode` Half of this function is now done in `merge.graft` and the over part is easy to redo manually. So we remove the usage of this function to remove the dependency of evolve on the rebase extension.

evolve: use `merge.graft` instead of `rebasenode` The core mechanism is properly factorised and have a more stable API. This is also a good step toward removing the dependency between evolve and the rebase extension. This also avoid a breakage with the future 3.3 where some of the rebase API changed.

readme: add a reference to the bugtracker in the contribute section Idea from Patrick Mezard

merge with stable

debian: update changelog to 5.0.1

test: adapt test to stable output Stable branch should pass test with mercurial stable branch.

Added tag 5.0.1 for changeset 0f3090785283

evolve: update in file version (I'll eventually -not- forget to do so)

Added tag 5.0.1 for changeset 51e1e962172c

readme: prepare for release 5.0.1

readme: update changelog

evolve: tell user which "base of divergent changeset" is not found

debugobsconvert: fix 6f08a8f6bb1d and add proper test The changeset adding deduplication (6f08a8f6bb1d) was super buggy and dropped all obsmarkers in the repo. We update the test to check this case and fix the deduplication.

merge with stable

debugobsconvert: fix a typo on "version"

debugobsconvert: also deduplicate markers in the process Various bug allowed identical or semantically identical marker to be added to the repository. So we also filter duplicates when converting.

evolve: fix capitalization in docstring for commands

test: ensure we echo prompt Tool running using the shipped 'run-tests.py' instead of the mercurial one would get different result otherwise.

debian: blacklist simple4server tests simple4server is not part of the package.

debian: blacklist test-drop drophack.py is not shipped.

debian: update control file to reflect dependency to mercurial 3.2

evolve: fix the version number The tag is 5.0.0 the content of the file was wrong.

help: fix typo in help for 'obsolete' option added to import command

test: add an explicit test for simple4server with bundle2 This makes it more likely to we catch issue with the iteration of these two.

Make next/prev only move bookmarks optionally

obsup: hack extension to make in-place upgrading of obsolete markers easy I tried upgrading my obsstore by doing a local clone as suggested by Pierre-Yves and Sean, but that caused me to end up with a ton of unstable changes that should have been marked dead. In fact, as far as I can tell, most of the performance win of that upgrade came from the fact that only about 46% (46672 of 102285) of my markers were brought over with the copy-and-pull method.

merge stable into default

evolve: fix selection of changeset to evolve from the middle of a stack (issue4434) The evolve algorithm picked a bad changesets to evolve (unstable changeset with no obsolete parent (but unstable parent). And then get confused assuming that is the first parent was not obsolete, the second should be. This changeset fixed the issue by preventing selection of direct descendant of [.] when evolving.

merge

merge back with compat branch

create a branch compatible with mercurial between 3.0.1 and 3.1.2

evolve: fix suggested commands in "parents are not common" hint The argument to 'hg rebase' is --dest, not -dest. Also, the same command should not be suggested twice -- the second option should be the rebasing the other way around.

evolve: add missing newlines in "parents are not common" hint

test-touch: add a test for rename preservation after touch This bug was recently fixed in 3.2.

evolve: fix error during iteration over bumped changesets (bug38) We are iterating over a manifest while altering it. We cannot use iteritems() in this context. we move to items, that will be an issue for very large changesets.

evolve: add various version info to save time on troubleshooting

readme: drop the mention to the wip repo The time spend in work in progress repo are fairly low nowaday.

readme: update the contribution process The inability to easily pull from pull request on bitbucket and the lack of obsolescence support there are too much pain. Patchbombed contribution are flowing so we drop the bitbucket mention in the README.

merge 5.x line into stable (drop pre hg-3.2 compat)

evolve: add the missing newline to output messages where needed

evolve: add the missing argument to a debug statement

uncommit: abort when rev specifies the current changeset The uncommit command now fails when the rev argument indicates the current changeset. Previously this aborted in the changeset obsolescence machinery when the current commit tries to obsolete itself. Using the rev argument like this indicates a misunderstanding, so printing an uncommit-specific error should make it easier to understand what went wrong.

test-amend: add a test for amend with no configured username (issue4211)

evolve: preserve branch change evolve will now preserve branch change. If the evolved changeset had a different phases than its parent, this branch name will be preserved. This allow propagation of branch change during amend but do not overwrite special branch name on children,

amend: allow the --logfile argument to work properly Though documented to work, using --logfile previously set the message option to that of the parent commit, and cause amend to abort with this message: abort: options --message and --logfile are mutually exclusive It's not clear why the 'edit' option is popped, but this option is only fetched in order to let the core commit command do the work.

Added tag 5.0.0 for changeset c13b408c0006

setup.py: bump version to 5.0.0

evolve: update tested with statement

README: update for 5.0 release

fixup previous changesets (after forgot to amend).

prune: stop reinjecting all selected revisions back into the revrange This trigger quadratic complexity for no good reason.

test: adapt to new hidden-changeset message From c1aede895072 we have a nicer error message when accessing public changesets: abort: hidden revision '6'! (use --hidden to access hidden revisions)

pullobsolete: handle migration from todosteps to donesteps This make evolve compatible with change made in Mercurial (92bf9abc4deb)

evolve: adapt to change in core rebase Mercurial core changeset 63e889cc610d (And the ones around it) changed the way graft and rebase work. We adapt to them.

test: ignore bytes lenght in message We should be storying floating point number as floating point number. So the size of the meta for fm0 format is going to be unstable. We cover them with a glob.

evolve: adapt to the new metadata storage Metadata are not stored as a "list" of (key, value) tuple instead of a binary blob. This adapt to ef880e28e56a in Mercurial.

debugobsstorestats: drop the average meta length Does not have much sense with the existence of multiple format with different meta storage. We drop it for now and will introduce and average markers length later.

debian: remove .PHONY line completely This does not work in the case of implicit pattern matching rules as used by dh. Consider this example from Geoffrey Thomas: $ ls -l total 16 -rw-r--r-- 1 faheem faheem 17 Aug 16 20:27 blue.c -rw-r--r-- 1 faheem faheem 17 Aug 16 20:27 green.c -rw-r--r-- 1 faheem faheem 35 Aug 16 20:26 Makefile -rw-r--r-- 1 faheem faheem 17 Aug 16 20:27 red.c $ cat Makefile %: %.c gcc -o $@ $< .PHONY: blue $ make red gcc -o red red.c $ make green gcc -o green green.c $ make blue make: Nothing to be done for blue'. The Make manual (http://www.gnu.org/software/make/manual/make.html#Phony-Targets) says Since it knows that phony targets do not name actual files that could be remade from other files, make skips the implicit rule search for phony targets (see Implicit Rules). Thanks to the good folks on #debian-mentors on OFTC for instruction and explanation in these matters.

debian: replace clean with override_dh_auto_clean The current version does not call the default dh_auto_build, which is a bug. This is required in particular for cleaning the build directories.

debian: replace build line with a override_dh_auto_build line This is more in line with how dh is supposed to work. In particular, dh_auto_build is functionally equivalent to the repeated "dh build --with python2 --buildsystem=python_distutils" used currently.

debian: remove clean target before implicit match rule This is redundant because it is a match-everything rule. Possibly a typo.

debian: Add DH_VERBOSE, commented out, for debugging convenience

evolve: wrap exchange.push() for compatability with core mercurial 4d52e6eb98ea

drophack: use `first` and `last` on smartset This matches the new changes in mercurial core.

evolve: use `first` and `last` on revset this match change in mercurial core. G: changed hgext/evolve.py

test: adapt to core mercurial 9ab18a912c44 This changesets improved the output of prompt during test.

evolve: adapt to core changeset 4f14303e8954 pushop.ret is now pushop.cgresult

merge with stable

evolve: update evolve merge message People are apparently unaware of --succ

import: fix the "is None" check for the Node We are checking the wrong variable.

test: remote "date" from the output of obsmarker meta Data is no longer a generic meta. it has its own field now.

test: adapt to mercurial changes Changeset 816be4ca4ae2 in core make hg config return non-zero when no value match the filter.

evolve: wrap setparents calls in begin/endparentchange Mercurial rev d259322a394b added a requirement to wrap any dirstate.setparents calls and their associated dirstate changes in begin/endparentchange.

merge with stable A couple of test fix were added on stable.

test: add glob annotations where Mercurial's run-tests.py suggests These happen to also be 5 of the 6 tests that warn about no result code when run on Windows. Probably not a coincidence. The tests previously warned about no result code whether or not it was run interactively. They only started working when test output changes were accepted in interactive mode, and these globs were added automatically.

evolve: really fix the 'grab' alias on Windows Something seems to have happened when importing cb36a4eb0157- it dropped the $@ argument to --rev, and it picked up a bunch of test changes for the obsolete sha1 replacement patch. Instead of backing it out and breaking the tests for the sha1 replacement, this just fixes the missing $@.

evolve: keep vague compatibility with 3.1 Matt Mackall work flow requires this to not crash in fire with 3.1. We apply basic bandage to stop the bleeding.

merge: all the stable sweetness into default

evolve: replace each obsolete sha1 in the description with its latest successor Obsolete csets are hidden by default and don't get pushed to the parent repo. In order to avoid broken references in commit messages, it makes sense to evolve those references to the latest and greatest successor, as each cset containing them is evolved. Of course, stale references can still occur if a commit in branch 'A' references something in branch 'B', and that something in 'B' is evolved but 'A' isn't subsequently evolved. This alleviates the user that is evolving a series of commits from having to 1) recognize that there is a hash that needs updating in any one of the series 2) look up the latest successor manually 3) hg amend -e The regular expression for matching and the logic for replacing are borrowed from the convert extension [1]. It might be nice for the output to state the reason that the reference couldn't be updated (it was pruned, split or diverged), but that may be excessive for something only displayed in verbose mode. (Maybe it should be a ui.status() instead?) [1] http://selenic.com/hg/rev/45562379ce4e

test: add parent hash to commit message in test-evolve.t This (slightly) reduces the changes to the test when hashes in the commit message are evolved in the next commit.

evolve: fix the 'grab' alias to work on Windows For some reason, the variable wasn't substituted on Windows in this case. From the test suite: $ hg grab 8 '' is not recognized as an internal or external command,\r (esc) operable program or batch file.\r (esc) This change seems hacky, but there isn't a readily available example of setting up a shell alias like this from a python module to know if there is a better way, and it seems like this is maybe just a convenience for tests, since there is no documentation for the aliases. The local copy of run-tests.py appears to predate the Windows support in the Mercurial repository's version, and fails each test with: The system cannot find the path specified. However, with this change and blacklisting test-simple4server.t, the tests can (mostly) be run on Windows like so: $ ../../hg/tests/run-tests.py --with-hg=../../hg/hg --blacklist windows Skipped test-simple4server.t: blacklisted Warned test-obsolete.t: no result code from test Warned test-tutorial.t: no result code from test Warned test-evolve.t: no result code from test Warned test-userguide.t: no result code from test Warned test-sharing.t: no result code from test Warned test-drop.t: no result code from test Failed test-prune.t: output changed # Ran 41 tests, 1 skipped, 6 warned, 1 failed.

debian: add README to the list of documents shipped with the package.

debian: add missing curl Build-Depends curl is required to run the tests.

prune: work around lazy revset slowdown Since 3.0 lazy revset is making some revset very slow. We currently work around the issue by using a simple loop instead.

readme: act that there will not be a 4.1.1 release

evolve: write a status message when the working directory changes Evolve changes the working directory during atop, etc which can confuse people as they are not aware that this happend. The ideal solution is to not have evolve change it, however mercurial limitations don't allow us to do that at the moment. So instead we are printing a status message that the working directory changed.

merge small default improvement before dropping compat with released Mercurial We are going to drop a bunch of compatibility layer.

pull: use discovery to pull less obsmarkers through bundle2 This hooks in the official API for bundle2

pull: run obsmarker discovery only on the common set If the discovery is run before the obsmarker push, pulled revs does not exist in the repository yet.

evolve: fix a typo in a long function name

bundle2: burn some more of the old code Some of it survived.

obsexch: respect todosteps when using the old push

compat: update the tests after the jump

compat: use in-core encodemarkers Now exist in core witha slightly different format.

compat: rename encodeonemarkers to _fm0encodemarkers This is getting renamed in core.

enable bundle2

drop the debug output

test-simple4server: add error output for http This is useful in case of failure.

compat: adapt to change to memctx Since core mercurial 650b5b6e75ed we must return None instead of raising IOError. Hooray for mads!!

evolve: use pushkeyescape from core It exists for all version we are compatible with.

evolve: display more message

evolve: drop a very old safe guard about bad marker The do not clean marker with nullrev successors. I assume none of them remains.

pull: drop the custom bundle2 part

obsexc: add support for evoext_pushobsmarkers_0 on localpeer This will unify the behavior of exchange in the tests.

evolve: allow read() from the magic StringIO

evolve: drop old obsmarker discovery code It as not caller anymore.

obsexc: use the common discovery We drop the hand made discovery and rely on the general one introduced some time ago.

obsexc: push nothing if push fail This is closer to the future behavior with bundle2.

discovery: hook on the official discovery for obsmarker This will let the discovery be done once and used by core.

discovery: use a different setup step for discovery This will ease the dropping of the other part later.

evolve: remove unused drop random

obshash: resist to unknown revision during discovery

code movement: gather discovery code together The discovery code is the one thing that will survive the great purge.

bundle2: drop custom compat We'll rely on the core implementation to use bundle2.

drop relevant marker compat

evolve: always use the transaction version of phase movement This drop compat with all known released version of Mercurial.

evolve: drop debugobsoleterelevant Core gained this capability usig `hg debugobsolete --rev REV`

evolve: yield to relevant markers handling in core The handling of obsolescence marker is now performed in core. If we detect a mercurial with such power we skip the wrapping from evolve. This changeset also suffer from massive diff from the change in display of obsolescence marker with parent information introduced at the same time in core.

test-wireproto: adapt to output change from core

test: update test to new date formatting in debuobsolete output The date is now a first class citizen with proper formatting.

test: update help output to future 3.2

merge with stable

pkg: make tests pass using the generated tarball Exclude tests/test-drop.t and include tests/_exc-util.sh, tests/dummyssh, and hgext/simple4server.py.

compat: always call metadata as a keyword argument Those function are gaining more arguments. using keyword argument keep use safe in all version.

test: work around a bug in debugobsolete The defautl date was wrong all along. So we use a fixed date for all tests.

merge with stable

uncommit: add a --rev argument The uncommit command now taks a --rev argument. This lets restore the file content in the current commit to another revision (instead of just is parent content). This still does not touch the working directory content.\ This allow the longly awaited: hg uncommit --hidden --rev 'precursors(.)'

readme: add missing changelog entry My dear friends changelog.

merge with stable

import: handle patch with no node information patch created by old Mercurial version or other tool have no node information. We cannot create marker in this case.

import: compare binary node with binary node when using --obsolete hex(N) != N

README: fix version number

relocate: fix typo

merge back with stable

Added tag 4.1.0 for changeset 4d5d101e878f

pkg: prepare release 4.1.0

merge default into stable Release 4.1.0 is coming

evolve: adds a --confirm option This option lets the use double check if it wants the evolve operation.

evolve: gather evolve output together We stop relying on ui.status being muted when --quiet is used. Instead we use plain write and hide this behind an explicite ui.quiet check. This will help use enforcing this output in the comming --confirm mode.

doc: forgotten amend

evolve: attempt to clarify the documentation around --all and --any This have been confusing to people. We try to be a bit more verbose in the evolve help.

evolve: re enable progress for all Progress is an important information for user facing slow push.

evolve: stop pretending we are pulling markers when we don't If discovery did not found any unknown marker, we should not display the "pulling marker" message.

evolve: do not even try to pull if all remote markers are known previously we were explicitly pulling an empty bundle.

evolve: add "obsolescence" in front of marked in the previous message

evolve: make the "X markers added" message during pull important

evolve: make the "pulling obsolescence marker" message important This is likely to take some time the user needs to know what is going on.

evolve: add "absolescence" in front of markers in the message

evolve: re-introduce the "pushing 42 markers…" message Informing the user about marker behind pushed sounds a good idea.

evolve: hide the OBSEXC message behind a config option This will lets most people ignore them while keep evolve dev having a close look at them. Some of the most useful messages will be reintroduced for all in coming changeset.

evolve: move OBSEXC progress in a dedicated function This will help use to rework them.

evolve: move OBSEXC message in a dedicated function This will help use to rework them.

merge with stable

Added tag 4.0.1 for changeset e914884fb721

evolve: prepare release 4.0.1

evolve: fix distribution for 4.0.0

evolve: update readme

evolve: update tested with statement

test: apply message changes introduced by mercurial 3.1

test: use `hg log` in the `getid` function using `hg id` to retrieve full lenght node requires to use --debug. This may include debug output (regarding branch cache invalidation for example) that breaks the test. We now use `hg log` with a custom template. This is a simple, clean and robust solution.

evolve: add a tr arguement to retractboundary Mercurial core is adding a transaction argument to its phase movement function. We do the same. If Mercurial appears to be an old version (without the extra args) we call it the old way.

relocate: add transaction around phase movement cheap way to acces access to the transaction object

uncommit tr

evolve: saner locking an transaction in `hg evolve` Each trouble solved used to handle locking and transaction on its own. We now have a top level locking and transaction. This will helps use making sure phase are moved within a transaction.

uncommit: saner locking scheme (and do it in the right order in the process)

prune: lock inside the try tss…

debugrecordpruneparents: release tr the same way than the lock Nicer code is nicer.

evolve: drop useless wlock in rewrite

divergence: actually use the hint that have been around for age Having a hint in the code is good, having it displayed is usually seens as better.

evolve: rename `tr` variable (for troubled) to `tro` tro is usually dedicated to transaction.

evolve: add a custom layer around retractboundary retract boundary is about to gain a new argument in core, we need somethign to handle this

merge with stable

evolve: point to the core mercurial tracker for bugfix As per Matt Mackall request evolution related bug should be filled again the main tracker as at some point al lof evolve wil be in core.

evolve: drop reference to 2.3 in the extension help text We are far away from 2.3 now.

readme: update the work in progress repository location

readme: update the official documentation location

evolve: deprecated --obsolete and --old-obsolete option on commit and graft Such option are unlikely to make it into core, so we deprecate them to avoid uer discovery them and getting used to them.

push: put phase in the same bundle2 than changegroup and obsmarkers Note that the usual phase push from core will be performed in all case. But the discovery should find already in sync phase at that time. The old ways sync will be properly skipped when this logic is introduced into core.

push: put obsmarkers in the same bundle2 than changeset When client is Mercurial 3.1 and server have proper evolve version and both side have bundle2 enabled, we'll includes a part containing obsolescence markers. When obsmarkers are included to the main bundle2 push, they will failed to be pushed if the changeset failed to be pushed (and reciprocally).

push: extract obsmarkers discovery in a dedicated function The need to factorise the computation of the markers to send in a dedicated function. This prepare the inclusion of obsmarkers in an atomic bundle2 push. For now, when obsmarkers are pushed using bundle2, they are still doing it on their own, in a dedicated bundle2 push.

evolve: update the tested with statement Tested with the soon to be release 3.1.

test: adapt to upstream message change A message gained parenthesis in 3.1 we reflect than in the test.

uncommit: improve error message for use with no arguments

test: use `hg log` in the `getid` function using `hg id` to retrieve full lenght node requires to use --debug. This may include debug output (regarding branch cache invalidation for example) that breaks the test. We now use `hg log` with a custom template. This is a simple, clean and robust solution.