sqlcache: passe better connection option These new options should help with handling transaction consistency and database access on high load.

contrib: introduce a small script to stress tests obsolescence exchange Tester have been reporting some error that seems to originate from concurrent access/update to the cache. This script will be useful to reproduce these situations locally.

changelog: mention the database robutness fix in the changelog

sqlcache: also catch malformed database error This is apparently another way for sqlite to fail at concurrency.

stablerange: rework saving logic to be clearer and safer

obshashrange: rework saving branching to be clearer and safer

sqlcache: protect read query too Some error (like locked database) can even happens when doing readonly operation. So we protect them too. At that point, it seems that pysqlite3 is not the right tool for this job.

sqlite: fast path when nothing to save If we have no new data to save, we should try to do anything. Doing something concurrently is hard enough. If we can avoid doing it at all that would be great.

sqlcache: initialize meta table last It turns out that pysqlite commit after each table creation. This commit is an hopeless attemps to make the caches more concurrent friendly.

obshashrange: always check in base cachekey against the recorded one Ignoring the check in the none case is just disaster waiting to happens.

stablerangecache: fix output in the drifted case The node were not hex, and the field had different order in the pair. This is now fixed.

sqlcache: also ignore integrity error integrity error can happens when multiple client tries to warm similar cache. Given we leave multiple hole in the cache that can be warmed by anyone, this is harder to avoid that we would like. We simply catch the error in this case. Someone will warm the missing entry later.

sqlcache: cache OperationError when saving This is a cache, so we should not crash when trying to operate on it. OperationError can happens when the database is seen as readonly or locked.

sqlcache: passe better connection option These new options should help with handling transaction consistency and database access on high load.

contrib: introduce a small script to stress tests obsolescence exchange Tester have been reporting some error that seems to originate from concurrent access/update to the cache. This script will be useful to reproduce these situations locally.

test-compat: merge mercurial-4.4 into mercurial-4.3

test: adjust glob in previous changeset I forgot to glob the time.

branching: merge with stable

packaging: mark as development version This avoid confusion.

Added tag 8.1.1 for changeset e7abf863e113

packaging: prepare release 8.1.1 Topic is untouched so we keep the previous version number.

branching: merge with stable

test-compat: merge mercurial-4.5 into mercurial-4.4

test-compat: merge mercurial-4.6 into mercurial-4.5

test-compat: merge stable into mercurial-4.6

changelog: add a line about improved prune documentation

changelog: reorder content

prune: more human-friendly documentation The previous help string generally had a highly technical language. I've tried my best to address that.

prune: improve help first line The previous help string for prune didn't mention successors in the first line. Note: Pierre-Yves David changed the original "hidden" to "obsolete" as it is better concept to put forward. (a bit more specific and with more consequence than just "hidden"). We need a good `hg help obsolete/obsolescence` result. We'll try to get this in the next version.

prune: try to clarify the "multi" revisions case I hope this highlight the use case for all three options.

prune: move help about biject after the one about split and fold The --biject flag seems a less common use case.

prune: no longer consider dropping --biject Field testing show it save people one wrong call often enough that it is worth keeping.

prune: more human-friendly documentation for --biject

prune: use ``-s/--succ`` in documentation Extracted by Pierre-Yves David

branching: merge with stable

compat: import the module dagutil module in evolve Discovery code is core have been rewritten to no longer depends on dagutil and the module have been removed. This break evolve's obsmarker discovery (older version of the discovery use the same concept).. A quick work around is to import the dagutil module in evolve and use that for now.

evolve: delete evolvestate after aborting interrupted `next --evolve` We didn't use to delete the evolvestate after aborting an interrupted `next --evolve`. This patch makes sure we do that. The test changes demonstrate the fix and we add a CHANGELOG entry about the fix.

tests: check that we delete evolvestate correctly after abort/stop This patch adds test to check whether we delete the evolvestate correctly after aborting or stopping an interrupted evolve which is caused because of `next --evolve`.

changelog: update with issue5927 fix

evolve: set the second parent if we get conflicts while evolving This patch makes sure we set the second parent correctly if we run into conflicts while evolving. Also, we need to make sure to reset to one parent before committing.

evolvecmd: import mercurial.node module as nodemod We use node as a variable name at few places and the conflicts with variable name are not good. Moreover this practice is followed in core mercurial also.

tests: demonstrate the interrupted evolve does not set p2 This patch adds a test to demonstrate issue5927 which is about evolving not setting p2 correctly in interruption.

cleanup: remove unused import

cleanup: remove old comment that lost all meaning

template: remove compatibility layer for `showobsolete` The keyword exist in core since 4.2

exthelper: drop compatibility layer for `command` We dropped support for 4.2.

topic: drop compat layer for `command` We dropped support for 4.2

legacy: drop compat layer for "command"

drophack: drop compat layer for "command"

compat: drop compatibility layer for updatecaches The `repo.updatecaches` method is support since Mercurial 4.2

branching: merge with stable

obshashrange: force saving of stablesort and firstmerge cache For some reason the check in repo is not working for stable sort. We also force firstmerge to be saved as test show it was missing.

test: add a test about cache warming This test reveal that multiple cache are not saved to disk after discovery (firstmerge, stablesort).

stablerange: save stablesort cache alongside the stablerange one This one does not introduce inconsistency but helps performance. G: changed hgext3rd/evolve/stablerangecache.py

obshashrange: always save stable range cache alongside the obshashrange one If the on disk date cover different area, the invalidation of affected range will misbehave.

obshashrange: clear in-memory cache alongside the sqlite one Clearing on disk data in nice, but we also need to remove in memory one or we'll get incorrect results.

branching: merge with fixes on stable

changelog: summarize the recent improvement

obshashrange: re-enabled more selective pruning of affected range

obshashrange: fix computation of affected ranges The computation of impacted nodes and associated revs is fully reworked. In addition,we introduce multiple new tests covering cases that were previous wrongly handled.

obshashrange: add more validation output to tests We are about to update the logic and add more tests.

obshashrange: correctly detect changeset directly affected by prune Before this changesets, parent of standard obsmarkers were wrongly considered affected and pruned changeset were wrongly not considered affected.

obshashrange: do not search for affected ranges above the highest we have It is a fast way to know we don't have an affected range for an affected revision.

obshashrange: do not search for affected stable range cache is unavailable Before this changeset we where resetting in all cases, and then looking for affected ranges. In addition is the stable range were not available, the functions silently returned no ranges affected. Now, do one or the other depending of the availability of the stable range cache. In practice we always do a broad reset because the code detecting affected changeset is currently buggy.

stablerange: build closure a bit less inefficiently The new way make me a bit less sad than the old one.

discovery: make sure repository wrapping happens in the right order Otherwise we may end up in situation were cache are not warmed in the right order, crashing.

branching: merge with stable

changelog: add an entry about the next --evolve fix

changelog: update 8.1.0 release date

next: fix topic constraint in the --evolve case We were using the wrong variable.

next: use set for membership testing There are no reason to use a list here.

test: adapt to core output change on unknown command CORE-TEST-OUTPUT-UPDATE: 5199c5b6fd29

compat: use new style template keyword declaration

test: fix previous changeset output The previous changeset got published prematurely.

obshashrange: add a test for `hg debugupdatecache` It seem to be properly filling the caches.

branching: merge with stable

templatekw: do not override in-core {obsolete} keyword The extension version uses old API, which will be deprecated in Mercurial 4.8.

packaging: mark version as development version This avoid confusion when installing non-tagged version.

Added tag 8.1.0 for changeset 2c5d79c6459c

packaging: prepare version

branching: merge stable into default

test-compat: merge mercurial-4.4 into mercurial-4.3

test-compat: merge mercurial-4.5 into mercurial-4.4

test-compat: merge mercurial-4.6 into mercurial-4.5

test-compat: merge stable into mercurial-4.6

compat: fix obslog compatiblity with 4.3 Due to the way of how successor sets are retrieved in new Mercurial version, retrieving it again in 4.3 way would be a performance hit and complexify the code too much. Don't retrieve markers of successor sets which makes obslog filtered output more limited in 4.3.

compat: add a fixed version of mergecopies for hg<=4.3 In hg<=4.3, copies._fullcopytracing() does not exist and we have to wrap copies.mergecopies() to fix the AssertionError coming from core.

test-compat: merge mercurial-4.5 into mercurial-4.4

test-compat: merge mercurial-4.6 into mercurial-4.5

test-compat: merge stable into mercurial-4.6

compat: make compatible call to ui.edit for hg <= 4.3 This is similar to what is already happening elsewhere.

compat: ignore extra status feature for hg <= 4.3 The feature does not exist in 4.3 and below.

format: fix indentation level Previous change got pushed without the fix by mistake.

test-compat: merge mercurial-4.6 into mercurial-4.5

test-compat: merge stable into mercurial-4.6

compat: disable empty topic warning on `hg phases` for hg <=4.5 This is annoying to support for lower version.

compat: restore compatibility with Mercurial <= 4.5 The __init__ arguments changed a lot in 4.6 and 4.7, lets do be as transparent as possible

test-compat: merge stable into mercurial-4.6

test-compat: merge stable into mercurial-4.6

branching: merge stable back into default

compat: improve detection of Mercurial 4.6 fixed case where AttributeError can be thrown when importing wireproto

cleanup: use NotImplementedError instead of NotImplemented Caught by pyflakes.

CHANGELOG: add some entries and remove duplicated entries

branching: merge default into stable The stable branch of Mercurial core now contains Mercurial 4.7 so evolve branch policy requires this merge. The @ bookmark is in the right location, so people doing clone will get to the latest release.

branching: merge with the other head

test-compat: backed out changeset 8fe8152ff071 These output change were for 4.7 only.

test-compat: backed out changeset 41fc764bf28c These output change are for mercurial-4.7 only

test-compat: backed out changeset dda5b2134b32 These change were for Mercurial 4.7 only.

test-compat: backout change to 1e103c7f7663

compat: use older API for older version This API changed in 4.7, we need to adjust.

branching: create mercurial-4.6 branch for updated test output This branch will be used to check for 4.6 compatibility.

builddependencies: extract function for getting successors The code no longer makes a difference between single- and mult-successor cases and just needs a set (or generator, really) of successors to iterate over. Let's extract that into a function.

builddependencies: share code between single- and multi-successor cases The two cases now have more similar structure ("if succ in revs: dependencies[r].add(succ)"), so let's share the code so it doesn't start drifting apart again.

builddependencies: build inverse dict from forward dict It's error-prone to keep "dependencies" and "rdependencies" in sync (we don't do it correctly when there are multiple successors or a node). It's easier to just create "rependencies" from "dependencies" after it's complete.

builddependencies: remove a use of defaultdict I don't see much reason to make "rdependencies" be a defaultdict when "dependencies" is not. It's easy to initialize each entry ourselves.

branching: merge with stable

builddependencies: don't remove found deps when divergence found (issue5946) It seems obviously wrong to not keep any dependencies for a revision that we had already found (for p1) if we run into divergence (from p2). It also happens to fix issue5946 :)

builddependencies: don't add dependency on revision outside input set This was already handled in the single-successor case, but had been missed in the multiple-successors case. Note that there seems to be a copy of builddependencies() in the topics extension. I don't use topics more than I have to, so I'll let someone else fix that code.

builddependencies: consider all divergent successors We were only considering one. In test-evolve-abort-conentdiv.t:165, the input revs were {5, 8} and dependency dict was {8: set([]), 5: set([10])}, which is a little weird (10 is not in the input set). It still worked because the callers used it (they seemed to only care if there were *any* dependencies). This patch fixes the issue by considering all successors. That means the dependency dict will be {8: set([]), 5: set([8, 10])} after this patch. The next patch will remove the 10 from that set.

tests: add test for issue5946 The test case is a copy of the one for 5833. The only difference is that the merge parents are recorded in the opposite order (and that the test is truncated because it fails).