refactoring: use unlinkpath with ignoremissing

docs: "deprecate" checkchanged and checkconflicts in merge-tool configuration These settings were replaced by check=changed and check=conflicts in a912f26777d3. There is no reason to announce two different ways to achieve the same. The old way should be kept but not announced.

tests: run with PYTHONHASHSEED=random Python set and dict iteration order is in principle undefined but usually 'quite stable'. Setting PYTHONHASHSEED=random will make the iteration order more random in Python 2.6.8 and 2.7.3 and where it has been backported. This can thus help spot dependencies on undefined behaviour and prevent future problems.

run-tests: exit cleanly if parallel run is interrupted If interrupted while running with "--jobs N", run-tests asynchronously spewed a bunch of output and backtraces from both the master and slave processes, leaving the terminal full of goop. This patch makes it behave more sensibly.

filtering: rename filters to their antonyms Now that changelog filtering is in place, it's become evident that naming the filters according to the set of revs _not_ included in the filtered changelog is confusing. This is especially evident in the collaborative branch cache scheme. This changes the names of the filters to reflect the revs that _are_ included: hidden -> visible unserved -> served mutable -> immutable impactable -> base repoview.filteredrevs is renamed to filterrevs, so that callers read a bit more sensibly, e.g.: filterrevs('visible') # filter revs according to what's visible

clone: don't include empty revlogs in stream

serve: don't send any content headers with 304 responses Fixes HTTP protocol violation introduced in cf5c76017e11. 'hg serve' would show a stacktrace when loading pages that not had been modified. There was test coverage for this, but the wrong response headers wasn't shown and thus not detected.

tests: fix doctest stability over Python versions pprint ain't pretty in Python 2.4: Changed in version 2.5: Dictionaries are sorted by key before the display is computed; before 2.5, a dictionary was sorted only if its display required more than one line, although that wasn’t documented. Fixes issue introduced in 404feac78b8a.

tests: stabilize doctest output Avoid dependencies to dict iteration order.

tests: make test-hook.t output more stable

convert: make toposort order stable when multiple orderings are possible The output of some tests are changed. Not necessarily to the better, just other valid permutations.

convert: report cvsps branchpoints sorted

convert: process subversion branch in a sorted order

convert: process missing branches in sorted order

convert: process splicemap in sorted order

mq: stabilize update after strip of parent revision Strip will (if it updates) update to the parent of revs[0], where revs are the roots of the tree that is stripped. When revs was list(set) it was thus undefined which root parent it would update to. With sorted(set) it is at least stable what it updates to. (But it is very possible that another more useful and predictable behaviour could be defined ... such as updating to the tip-most surviving wd ancestor.)

histedit: process bookmarks in sorted order

churn: sort users with same churn by name This makes the output order well-defined and improves code readability.

largefiles: upload files in sorted order

hgweb: generate query strings with parameters sorted by key

identity: report bookmarks sorted

obsolete: process markers in a stable order Using set iteration order gave unstable debugsuccessorssets output with PYTHONHASHSEED=random.

subrepos: process subrepos in sorted order Add sorted() in places found by testing with PYTHONHASHSEED=random and code inspection. An alternative to sprinkling sorted() all over would be to change substate to a custom dict with sorted iterators...

bookmarks: process pulled remote bookmarks in sorted order

copies: report found copies sorted

discovery: process heads in sorted order

merge: process files in sorted order

commandserver: report capabilities sorted

bisect: store state sorted

localrepo: store branchheads sorted

localrepo: store requirements sorted

copies: make the loss in _backwardcopies more stable A couple of tests shows slightly more correct output. That is pure coincidence.

serve: use chunked encoding in hgweb responses 'hg serve' used to close connections when sending a response with unknown length ... such as a bundle or archive. Now chunked encoding will be used for responses with unknown length, and the connection do thus not have to be closed to indicate the end of the response. Chunked encoding is only used if the length is unknown, if the connection wouldn't be closed for other reasons, AND if it is a HTTP 1.1 request. This will not benefit other users of hgweb ... but it can serve as an example that it can be done.

serve: remove connection close hack for Python 2.3 Introduced in 276de216d2c5 and no longer needed - BaseHTTPServer handles connection closing just fine if we don't mess with its internals.

hgweb: pass the actual response body to request.response, not just the length This makes it less likely to send a response that doesn't match Content-Length.

hgweb: don't pass empty response chunks on hgweb internals will often produce empty writes - especially when returning compressed data. hgweb is no middleware application and there is thus no reason to pass them on to be processed in other layers.

hgweb: remove handling of any else than strings from request.write Iterators should be returned WSGI style, not written. And apparently all of hgweb do that.

serve: send response headers even if response has no body The headers would usually be sent anyway because the app did a number of writes of empty strings.

hgweb: simplify wsgirequest header handling Remove leaky header abstraction and prepare for other encodings.

hgweb: make type a mandatory parameter to request.respond There will thus always be headers and the runtime check can be removed.

hgweb: use Content-Length for pushres This prevents some unnecessary http connection close.

hgweb: send Content-Length 0 for zero length response Before, Content-Length wasn't sent for 0 length responses. Now it is. This could in principle prevent some unnecessary http connection close.

archival: tarit should never close the dest passed to it Some archive types closed the open file passed to it, some didn't. This could cause either missing or duplicate close and cause problems in hgweb. The fix in 14f3795a5ed7 should only have closed the compressors and archivers - not the underlying file itself if no compressor is used.

mq: checktoppatch should only check if p1 is qtip There is no way qtip in p2 could be used for anything, and there is thus no reason to check and accept it.

mq: fix qpop of working directory parent patch when not at qtip mq assumed that it had to update from qtip to qparent, and instead of updating from where it was it failed with: abort: working directory revision is not qtip

largefiles: make log match largefiles in the non-standin location too Yet another match hack.

log: make log work even if first parameter doesn't exist A slowpath optimization kicked in too often because of wrong indentation.

merge: remove "case" comments The comments introduced in f05c182430a0 seems important ... but the context has apparently been lost and they do not have any value now.

merge: merge file flags together with file content The 'x' flag and the 'l' flag are very different. It is usually not a problem to change the 'x' flag of a normal file independent of the content, but one does not simply change the type of a file to 'l' independent of the content. This removes the fmerge function that merged both 'x' and 'l' independent of content early in the merge process. This correctly introduces some conflicts instead of silent incorrect merges. 3-way flag merge will now be done in the resolve process, right next to file content merge. Conflicts can thus be resolved with (slightly inconvenient) resolve commands like 'resolve f --tool internal:other'. It thus brings us closer to be able to re-solve manifest merge after the merge and avoid prompts during merge. This also removes the "conflicting flags for a - (n)one, e(x)ec or sym(l)ink?" prompt that nobody could answer and that made it easy to mix symlink targets and file contents up. Instead it will give a file merge where a sufficiently clever merge tool can help resolving the issue.

tests: better test coverage of merges of flags This makes existing problems with merges of symlinks and files more obvious and add test coverage for tricky merges without real common ancestors.

merge: remove old pre-audit code checking for absolute paths Audit will handle this in a more elegant way.

merge: drop reference to file contents immediately after write Like 7d2aaeea67ed this reduces memory usage on large merges.

merge: changing the mode of a file is also an update It was a change in the file system that wasn't reported in the summaries.

merge: use util.unlinkpath for removing moved files - more like how removed files are removed.

merge: .hgsubstate is special as merge destination, not as merge source

merge: remove redundant unlink after merge The early prescan for move/remove and removal of moved files in applyupdates was introduced with mergestate 368a4ec603cc and rendered this chunk of code irrelevant. The impact of the chunk was reduced in 5b3383ea67d2 - but it could have been removed completely.

merge: rename list of actions from action to actions

merge: consistently use "x" instead of 'x' for internal action types This makes it simpler to search for places where the action types are handled.

merge: consistently use repo.wopener.audit instead of creating a new auditor

scmutil: simplify vfs.audit - drop wrapped vfs.auditor

util: copyfile: remove dest before copying This prevents spurious problems writing to locked files on Windows.

merge: warn when internal:merge cannot merge symlinks A follow-up to d084df89d948. internal:merge should never be picked for merging symlinks ... but in the test suite we have HGMERGE="internal:merge" which bypasses all the usual merge-tool cleverness. Without any output it can be hard to figure out what happened and where the problem is.

merge with stable

histedit: correct the number of added revisions in online help In the context of help document on which this patch focuses, the example repository, which is source of cloning, should be already histedit-ed, and contain only 3 revisions (rev # 0 to 2). So, not 3, but 4 revisions should be added to the destination repository of cloning, if it contains 7 revisions (rev # 0 to 6). This patch also adds modifier "histedit-ed" to "example repository", to make context clear.

histedit: correct changeset IDs in online help There is no '633536316234' revision in the example repository. It should be 'c561b4e977df', according to the revisions in it and explanation in help document.

convert: correct 'hooks' section name in online help The section name for hooks is not 'hook', but 'hooks'.

hgweb: document the revnavgen function

hgweb: no do not use listinsert(0, ...) This is not efficient. We now append element and either reverse the list or iterate in reverse order.

localrepo: drop unnecessary check on wlock unlock Calling out to dirstate one line before guarantees that it'll be in the filecache.

dirstate: refresh _branch cache entry after writing it

filecache: create an entry in _filecache when __set__ is called for a missing one Preserve the invariant that if P is a filecached property on X then P in X.__dict__ => P in X._filecache. Previously, it was possible for a filecached property to become out of sync with the filesystem if it was set before getting it first, since the initial filecacheentry was created in __get__. Old behaviour: repo.prop = x repo.invalidate() # prop has no entry in _filecache, it's not removed # from __dict__ repo.prop # returns x like before without checking with the # filesystem New: repo.prop = x # an empty entry is created in _filecache repo.invalidate() # prop is removed from __dict__ repo.prop # recreates prop

filecache: allow filecacheentry to be created without stating in __init__ Will be used for properties that are set without getting them first.

rollback: don't clear the filecache This was an old workaround to force the filecache to reload everything. Now that the syncing issue is fixed, we no longer need it.

destroyed: keep the filecache in sync with __dict__ (issue3335) (issue3693) (issue3743) We need to make sure that if X is in the filecache then it's also in the filecache owner's __dict__, otherwise it will go out of sync: repo.X # first access to X, records stat info in # filecache and updates __dict__ repo._filecache.clear() # removes X from _filecache but it's still in __dict__ repo.invalidate() # iterates over _filecache and removes entries # from __dict__, but X isn't in _filecache, so # it's kept in __dict__ repo.X # X is fetched from __dict__, bypassing the filecache

localrepo: write the phasecache when destroying nodes

localrepo: update the branchmap when destroying nodes This was previously called directly during strip. Moving it to destroying also means that it'll be called through _rollback, which seems harmless.

localrepo: introduce destroying function

localrepo: don't refresh filecache entries that aren't in __dict__ We call invalidate to remove properties from __dict__ because they're possibly outdated and we'd like to check for a new version. Next time the property is accessed the filecache mechanism checks the current stat info with the one recorded at the last time the property was read, if they're different it recreates the property. Previously we refreshed the stat info on all properties in the filecache when the lock is released, including properties that are missing from __dict__. This is a problem because: l = repo.lock() repo.P # stat info S for P is recorded in _filecache <changes are made to repo.P indirectly, e.g. underlying file is replaced> # P's new stat info = S' l.release() # filecache refreshes, records S' as P's stat info At this point our filecache contains P with stat info S', but P's version is from S, which is outdated. The above happens during _rollback and strip. Currently we're wiping the filecache and forcing everything to reload from scratch which works but isn't the right solution.

changelog: please check-code and remove tabs Tabs were introduced in 06185554e7e3.

branchmap: Save changectx creation during update The newly introduced `branchmap` function allows us to skip the creation of changectx objects. This speeds up the construction of the branchmap. On the mozilla repository (117293 changesets, 15490 mutable) Before: ! impactable 19.9 ! mutable 0.576 ! unserved 3.16 After: ! impactable 7.03 (2.8x faster) ! mutable 0.352 (1.6x) ! unserved 1.15 (2.7x) On the cpython repository (81418 changesets, 6418 mutable) Before: ! impactable 15.9 ! mutable 0.451 ! unserved 0.861 After: ! impactable 6.55 (2.4x faster) ! mutable 0.170 (2.6x faster) ! unserved 0.289 (2.9x faster) On the pypy repository (58852 changesets) Before: ! impactable 13.6 After: ! impactable 6.17 (2.2x faster) On my Mercurial repository (18295 changesets, 2210 mutable) Before: ! impactable 23.9 ! mutable 0.368 ! unserved 0.057 After: ! impactable 1.31 (18x faster) ! mutable 0.042 (8.7x) ! unserved 0.025 (2.2x)

changelog: add a `branch` method, bypassing changectx The only way to access the branch of a changeset is currently to create a changectx object and access its `branch()` method. Creating a new Python object is costly and has a huge impact on code doing heavy access to `branch()` (like branchmap). This change introduces a new method on changelog that allows direct access to the branch of a revision. See the next changeset for impact.

branchmap: pass revision insteads of changectx to the update function Creation of changectx objects is very slow, and they are not very useful. We are going to drop them. The first step is to change the function argument type.

perf: add perfbranchmap command The command times the update of a branchmap from its nearest subset or from scratch.

clfilter: enforce hidden filtering on all repository accesses We ensure all repositores created through `mercurial.hg.repository` are "hidden" filtered. This is an even stronger enforcement than 5bb610f87d1d. Citing Matt's response to changeset 5bb610f87d1d installing filtering in dispatch: > Unfortunately, this means that code that doesn't go through dispatch (ie all > those crazy misguided people using Mercurial as a library) are going to see > these hidden changesets. > > Might be better to instead install the filter in localrepo construction by > default and disable it in dispatch.

archival: avoid touching deprecated gzip name attribute The existing workaround didn't work when no filename was specified. If running in a context with warnings enabled and subsecond mtime it gave a warning: DeprecationWarning: use the name attribute

archival: pass integer to struct.pack int field instead of float If running in a context with warnings enabled and subsecond mtime it gave a warning: DeprecationWarning: integer argument expected, got float

largefiles: fix update from a merge with removed files A situation with this case could happen after interrupting an update. Update would fail with: abort: No such file or directory: $TESTTMP/f/.hglf/sub2/large6 Update from a merge without using clean is not possible anyway, so this patch takes a step in the right direction so it gets as far as reporting the right error.

largefiles: fix revert removing a largefile from a merge Before revert could fail with: abort: .hglf/large@33fdd332ec64: not found in manifest! The LookupError will now be caught and handled correctly.

largefiles: remove unused proto.refuseclient code Should have been removed with other pre-1.9 code in 7c604d8c7e83.

subrepo: make 'in subrepo' string easier to find by external tools This patch is meant to make it easier for tools that wrap the mercurial output (such as TortoiseHg) to find the "in subrepo MYSUBREPO" string that (since 9e3910db4e78) is appended after subrepo error messages, particularly when the mercurial output is translated to a non-English language. The message remains the same but the '%s' that was used to prepend the original error message in front of the 'in subrepo' string has been moved out of the translatable string. As an example of the usefulness of making it easy to look for "in subrepo MYSUBREPO" strings, TortoiseHg looks for these strings in error messages in order to "linkify them" (i.e. convert "MYSUBREPO" into alink to the corresponding subrepo). The original string made it hard for a tool such as TortoiseHg to look for the translated string on mercurial's output because the translated string contained the error message itself. This meant that a regular expression was required to ignore the error message part. With this change TortoiseHg can just get the translated "(in subrepo %s)" string, substitute %s for the subrepo path (which it gets from the subrepo exception) and simply search for the resulting string (no regular expression needed, or at least a much simpler regular expression could be used). Additionaly, the existing string could lead a translator mistakenly assume that it was possible invert the order of the %s (error and subrepo path) fields, which would not work because the string interpolation was position based.

subrepo: fix python2.4 compatibility after 9aa6bee6e9f9 super(SubrepoAbort, self).__init__(*args, **kw) raises TypeError: super() argument 1 must be type, not classobj

test-keyword: improve grammar and spelling in branchcache note See: da9e544c69d6

Merge with stable.

repoview: extract hideable revision computation in a dedicated function This will help extensions to plug into the hidden mechanism.

largefiles: make update with backup files in .hglf slightly less broken Largefiles update would try to copy f to f.orig if there was a .hglf/f.orig . That is in many many ways very very wrong, but it also caused an abort if f didn't exist. Now it only tries to copy f if it exists.

tests: fix up test-highlight for breadcrumb changes

color: add template label function

templater: add no-op template function 'label'

posix: fix split() for the case where the path is at the root of the filesystem posixpath.split() strips '/' from the dirname *unless it is the root*. This patch reproduces this behavior in posix.split(). The old behavior causes a crash when creating a file at the root of the repo with localrepo.wfile() when the repo is at the root of the filesystem.

record: remove unused import

cvsps: fix indentation

hghave: introduce a test (unused) for cvs >= 1.12

discovery: drop the visibleheads function It has no users left. (Using filtering explicitly is easier.)

clfilter: remove the last usage of `visibleheads` We can just inline the filtering now that all server code uses changelog filtering directly.

discovery: drop the visiblebranchmap function

clfilter: drop extra filtering in wireprotocol The repository used by wireprotocol is already filtered. We no longer need to call special functions.

clfilter: filter "unserved" on all wireprotocol command calls This ensures that unserved changesets are not exposed through the wire protocol.

clfilter: drop extra filtering in localpeer The repository used by localpeer is already filtered. We no longer need to call a special function.

clfilter: make localpeer use a repo with "unserved" filter This changeset installs a broad filter on most repos used for serving. This removes the need to use the `visiblehead`/`visiblebranchmap` functions, and ensures that changesets we should not serve are in fact never served. We do not use filtering on hgweb yet, as there is still a number of issues to solve there.

performance: speedup computation of extinct revisions In their current state, revset calls can be very costly, as we test predicates on the entire repository. This change drops revset calls in favor of direct testing of the phase of changesets. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! extinct ! wall 0.015124 After: ! extinct ! wall 0.009424 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! extinct ! wall 0.032844 After: ! extinct ! wall 0.000066

performance: speedup computation of suspended revisions In their current state, revset calls can be very costly, as we test predicates on the entire repository. This change drops a revset call in favor of direct testing of the phase of changesets. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! suspended ! wall 0.014319 After: ! suspended ! wall 0.009559 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! suspended ! wall 0.033373 After: ! suspended ! wall 0.000053

performance: speedup computation of unstable revisions In their current state, revset calls can be very costly, as we test predicates on the entire repository. This change drops revset call in favor of direct testing of the phase of changesets. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! unstable ! wall 0.017366 After this changes: ! unstable ! wall 0.008093 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! unstable ! wall 0.045190 After: ! unstable ! wall 0.000032

performance: speedup computation of mutable revisions In their current state, revset calls can be very costly, as we test predicates on the entire repository. The "mutable" filter is used during branch cache loading operation. We need to make it fast. This change drops revset calls in favor of direct testing of the phase of a changeset. Performance test on my Mercurial checkout - 19857 total changesets, - 1646 mutable revision Before: ! mutable ! wall 0.032405 After: ! mutable ! wall 0.001469 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 mutable changeset, Before: ! mutable ! wall 0.188636 After: ! mutable ! wall 0.000022

performance: speedup computation of unserved revisions In their current state, revset calls can be very costly, as we test predicates on the entire repository. The "unserved" filter is used in multiple applications, and in particular in some branch cache loading operations. We need to make it fast. This change drops revset calls in favor of direct testing of the phase of a changeset. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! unserved ! wall 0.030477 After: ! unserved ! wall 0.011844 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! unserved ! wall 0.111259 After: ! unserved ! wall 0.000084

performance: speedup computation of hidden revisions In their current state, revset calls can be very costlys, as we test predicates on the entire repository. The hidden filter is very widely used, and needs to be very fast. This change drops revset calls in favor of direct revision manipulation. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! hidden ! wall 0.077553 After this changes: ! hidden ! wall 0.011230 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! hidden ! wall 0.389472 After: ! hidden ! wall 0.000079

performance: speedup computation of obsolete revisions In their current state, revset calls can be very costly as we test predicates on the entire repository. As obsolete computation is used by the "hidden" filter, it needs to be very fast. This changet drops the revset call in favor of direct testing of the phase of a changeset. Performance test on my Mercurial checkout - 19857 total changesets, - 1584 obsolete changesets, - 13310 obsolescence markers. Before: ! obsolete ! wall 0.047041 After: ! obsolete ! wall 0.004590 Performance test on a Mozilla central checkout: - 117293 total changesets, - 1 obsolete changeset, - 1 obsolescence marker. Before: ! obsolete ! wall 0.001539 After: ! obsolete ! wall 0.000017

clfilter: drop unnecessary explicit filtering on histedit Hidden changeset filtering is now done at repo level. The orphaned children computation will not include any (unless you add --hidden).

clfilter: drop unnecessary explicit filtering on rebase Hidden changeset filtering is now done at repo level. The rebaseset computation will not include any (unless you add --hidden).

clfilter: ensure that hidden filtering is working on all commands Now that hidden changeset are filtered for all commands, we test the behavior of `heads` and `summary` regarding hidden changeset.

clfilter: enforce hidden changeset globally The dispatch code now enables filtering of "hidden" changesets globally. The filter is installed before command and extension invocation. The `--hidden` switch is now global and disables this filtering for any command. Code in log dedicated to changeset exclusion is removed as this global filtering has the same effect.

record: use patch.diffopts to account for user diffopts This allows user defined diff options (e.g. showfunc) to be accounted for when using record. A test has been updated accordingly.

convert: fix most test-check-code-hg violations in cvsps code

tests: update hgweb tests to include breadcrumbs

subrepo: add subrepo property to SubrepoAbort exceptions This new property contains the path of the subrepo that generated the exception. This information can then be used by GUI tools such as TortoiseHg.

tests: update test-convert-cvs*.t The preceding commit caused their outputs to change.

cvsps: use commitids (when present) to detect changesets Simplify core logic by no longer attempting to work around missing class attributes. Instead always generate the attributes and ignore the cache if the attributes are missing

hgweb, spartan: link from manifest title to changeset page

hgweb, spartan: add "URL breadcrumbs" This change adds a "URL breadcrumb" to the "title" of the pages on the spartan template. By title I mean the first line that is shown right below the page selection row, which shows the name of the page that is being viewed, along with some additional information. In doing so it standarizes those "titles" which now follow the pattern: URL breadcumb / page details

hgweb: add a "URL breadcrumb" to the index and repository pages The purpose of this change is to make it much easier to navigate up the repository tree when the hg web server is used to serve more than one repository. A "URL breadcrumb" is a path where each of the path items can be clicked to go to the corresponding path page. This lets you go up the folder hierarchy very quickly. For example, when showing the list of repositories in http://myserver/myteams/myprojects, the following "breadcrumb" will be shown: Mercurial > myteams > myprojects Clicking on "myprojects" reloads the page. Clicking on "myteams" goes up one folder. Clicking on the leftmost "Mercurial" goes to the server root. This "breadcrumb" also appears on all repository pages. For example on the summary page of the repository at http://myserver/myteams/myprojects/myrepo the following will be shown: Mercurial > myteams > myprojects > myrepo / summary This change has been applied to all templates that already had a link to the main repository page (i.e. gitweb, monoblue, paper and coal) plus to the index page of the spartan template. In order to make the breadcumb links stand out the some of the template styles have been customized.

merge: never do premerge on symlinks Simplemerge is not symlink aware and will never do the right thing on symlinks.

merge: make internal merge fail cleanly on symlinks Simplemerge is not symlink aware and will never do the the right thing on symlinks. It would read the symlink as a file and abort with 'No such file or directory' on dangling symlinks. Instead, internal:merge now simply fails to merge symlinks.

debugpushkey: list keys sorted

debugdiscovery: report heads in sorted order

hidden: drop of the repo.hiddenrevs property It does not have any user left

context: retrieve hidden from filteredrevs This prepare the dropping of the repo.hiddenrevs property

revset: retrieve hidden from filteredrevs This prepare the dropping of the `repo.hiddenrevs` property

hidden: use both parents of working directory If we are merging with and extinct revision, this extinct revision should not be hidden.

hidden: drop cache on hiddenrevs property The `filteredrevs` function already have a cache mechanism. And this cache in invalidated at the same time than the current property cache. So we drop the cache on the property. The property itself is going to be dropped soon.

hidden: move computation in filter function There is not good reason for this computation to be handle in a different way from the other. We are moving the computation of hidden revs in the filter function. In later changesets, code that access to `repo.hiddenrevs` will be updated and the property dropped.

branchcache: add note about cache invalidation to test-keyword.t [Should've been included in aff706b3a21c.] --Kevin Bullock <kbullock@ringworld.org>

clfilter: add impactable filter The `mutable` filter still have some chance to get invalidated. This will happen when: - you garbage collect hidden changeset, - public phase is moved backward, - something is changed in the filtering (this could be fixed) So we introduce an even more stable filtering set: everything with a revision number egal or higher than the first mutable changeset is filtered. The only official use of this filter is for branchcache.

clfilter: add mutable filtering It filters all mutable changesets, leaving only public changeset unfiltered. This filtering set is expected to be much more stable that the previous one as public changeset are unlikely to disapear. The only official use of this filter is for branchcache.

run-tests.py: fix handling of symlink to the right python Before: a symlink for python in BINDIR was sometimes created, but it was never updated when a different Python was used and it was never removed. An invalid python could thus be left around and used when testing with --local. Now: the symlink is removed when wrong and created when necessary. The mechanism for finding the right name (python or python.exe) also had to be simplified and made more explicit.

log: use "hidden" filtering instead of manual check at display time When log is not given the --hidden option, hidden revision are not shown. We move the implementation from manual checking at display time to changelog filtering. This is the first official usage of the hidden filtering.

clfilter: introduces a hidden filter This filter exclude all hidden revision. We plan to use this filter to hide revision instead of manually checking contents of the hidden revisions set.

perftest: allow selection of volatile set to benchmark This helps when you focus on a subset of the volatile chain.

perftest: add a command to benchmark construction of volatile cache Obsolescence and filtering related caches are critical. Having a handy way to check them is valuable.

perftest: add an option to invalidate volatile cache Some revsets are sensitive to such initialization. Being able to test the impact is great.

perftest: document the perfrevset command I'll add an argument to it.

perftest: migrate to new style command declaration Declaring synopsis and argument on the function site is much clearer.

perftest: drop duplicated codes The `perfnodelookup` lookup commands is duplicated. We drop the first version, overwritten by the seconds.

cmdutil: make options argument optional There is not reason to force passing of an empty options list.

branchmap: allow to use cache of subset Filtered repository are *subset* of unfiltered repository. This means that a filtered branchmap could be use to compute the unfiltered version. And filtered version happen to be subset of each other: - "all() - unserved()" is a subset of "all() - hidden()" - "all() - hidden()" is a subset of "all()" This means that branchmap with "unfiltered" filter can be used as a base for "hidden" branchmap that itself could be used as a base for unfiltered branchmap. unserved < hidden < None This changeset implements this mechanism. If the on disk branchcache is not valid we use the branchcache of the nearest subset as base instead of computing it from scratch. Such fallback can be cascaded multiple time is necessary. Note that both "hidden" and "unserved" set are a bit volatile. We will add more stable filtering in next changesets. This changeset enables collaboration between no filtering and "unserved" filtering. Fixing performance regression introduced by 47f00b0de337

repoview: add a subset table This will be used by branchmap collaboration. See inline documentation for more details

branchmap: add a copy method If we want branchcache of different filter to collaborate, they need a simple way to copy each other. This will ensure that each filtered have no side effect on other filter level cache.

clfilter: use empty frozenset intead of empty tuple This will allows set operation needed for cache collaboration.

run-tests.py: don't use console for stdin when running in debug mode Tests would wait for input instead of using non-interactive mode.

tests: make hghave and run-tests exit on unknown feature requirements

test-largefiles.t: fix wrong '#if hgweb' - it should be '#if serve' A bug introduced in 0c1d10351869 did that a part of the test never was run.

dispatch: show empty filename in OSError aborts Mercurial would sometimes exit with: abort: No such file or directory where str of the actual OSError exception was the more helpful: [Errno 2] No such file or directory: '' The exception will now always show the filename and quote it: abort: No such file or directory: ''

test-dispatch.t: remove incorrect "cd $dir" A line that should have been removed in 50fbe9063ff2.

merge with crew-stable

dispatch: handle empty `testedwith` value in extension When extensions had an empty `testedwith` attribute the code tried to parse it and failed. As a result the actual error were shallowed by a This crash. We now treat empty strip as 'unknown'

destroyed: filter unknown before computing branchcache Branchcache of filtered version need up to date phase data.

branchmap: disable fallback to unfiltered branchcache Disables this simple optimisation to allow coming more powerfull approach: cache collaboration. Our goal is to have branchcache collaborate. This means that unfiltered branchcache will fallback to some filtered branchcache if invalid. We can't have the filtered branchcache to use the unfiltered one. That would loop.

localrepo: filter unknown nodes from the phasecache on destroyed When commit is followed by strip (qrefresh), phasecache contains nodes that were removed from the changelog. Since phasecache is filecached with .hg/store/phaseroots which doesn't change as a result of stripping, we have to filter it manually. If we don't write it immediately, the next time it is read from disk the nodes will be filtered again. That's what happened before, but there's no reason not to write it immediately. The change in test-keyword.t is caused by the above.

phases: make _filterunknown a member function of phasecache We'd like the ability to call filterunknown on an existing phasecache instance after nodes are destroyed.

localrepo: drop `_cacheabletip` method It iss dead code now.

branchmap: drop `_cacheabletip` usage in `updatecache` Nobody overwrite the `_cacheabletip` any more. We always update the cache for the whole repo and write it to disk (or at list try to). The `updatecache` code is simplied to remove the double phase logic associated with _cacheabletip.

mq: drop `_cacheabletip` usage Strip have dedicated work around to solve the same problem, strip is even a fraction faster without that thanks to simpler update process of the branchcache.

bundlerepo: drop use of `_cacheabletip` Now that bundlerepo use a read only VFS, we do not worry about what part of the branchmap is written back to disk. Nothing is written at all.

bundlerepo: enforce reading from core repo only We do not want anything computed with the bundle overlay to be written back in the repo. Such write will likely contains invalid data. The short terms goal of this change is to drop use of `_cacheabletip` in bundle repo.

branchmap: ignore Abort error while writing cache Read only vfs can now raise Abort exception. Note that encoding.local are also a possible raiser.

vfs: add a read only vfs This read only wrapper is intended to be used bundle repo. See follow up commit for details.

branchmap: read return None in case of failure This makes a clear distinction between having read a valid cache on disk or not. This will help caches of various filtering level to collaborate.

clfilter: fallback to unfiltered version when linkrev point to filtered history On `filectx`, linkrev may point to any revision in the repository. When the repository is filtered this may lead to `filectx` trying to build `changectx` for filtered revision. In such case we fallback to creating `changectx` on the unfiltered version of the reposition. This fallback should not be an issue because `changectx` from `filectx` are not used in complex operation that care about filtering. It is complicated to work around the issue in a clearer way as code raising such `filectx` rarely have access to the repository directly. Linkrevs create a lot of issue with filtering. It is stored in revlog entry at creation time and never changed. Nothing prevent the changeset revision pointed to become filtered. Several bogus behavior emerge from such situation. Those bugs are complex to solve and not part of the current effort to install filtering. This changeset is simple hack that prevent plain crash in favor on minor misbehavior without visible effect. This "hack" is longly documented in to code itself to help people that would look at it in the future.

phases: prepare phase command for filtering The phase command have some logic to report change made. We ensure this logic run unfiltered. With --force the command can change phase of a changeset for public to draft. Such change can lead to obsolescence marker to apply again and the changeset to be "hidden". If we do not run the logic unfiltered it could failed to fetch the phase of a newly filtered changeset.

phases: avoid changectx creation while checking command result This minor changesets saves the creation of a `changectx` ctx object only used to fetch the revision number.

test: use obsolescence marker to test hidden Instead of using a custom dedicated extension.

test-command-template.t: make "age" filter test work on Feb 29th reported by Julien Cristau.

merge with stable

Added signature for changeset a4765077b65e

Added tag 2.4.2 for changeset a4765077b65e

merge with i18n

merge i18n heads

hgweb, paper: add (Atom) subscribe links to the repository index This is similar to the subscribe links that already exist in other templates. Rather than the usual RSS and Atom links a single feed icon linking to the atom-log is shown.

hgweb: add (Atom) subscribe link to the main paper template pages The subscribe link is found at the bottom of the navigation sidebar. This uses a free icon from http://feedicons.com.

test-command-template.t: fix test so it all year This test started failing for me after midnight UTC on December 31st. Fixed it by specifying a date 7 years in the future more precisely (rather than just adding 8 to the year and specifying January 1st), which allows the test to pass both now and on 2012-12-01 at the same time.

amend: prevent loss of bookmark on failed amend The active bookmark were moved to the temporary commit. When the transaction were rollbacked, the bookmark were lost. We now temporarly disable the bookmark to prevent this effect.

amend: invalidate dirstate in case of failure (issue3670) The temporary commit created by amend update the dirstate. If the final commit fails, we need to invalidate the change made to the dirstate, otherwise the release of the wlock will write the dirstate created after the rollbacked temporary commit. This dirstate writing logic should probably be handled in the same object than the transaction one. However such change are too big for stable.

i18n-ru: delete loose letter

i18n-ru: synchonized with 777084ac8416

paper: sanity-check page feed links filelog has feeds in header, but not in menu bar help has header feeds pointing to tags

zeroconf: use port from server instead of picking port from config (issue3746) In order to get the port, wrap create server instead of hgweb_mod/hgwebdir_mod.

scmutil: don't try to match modes on filesystems without modes (issue3740)

hgwebdir: honor web.templates and web.static for static files (issue3734)

zeroconf: use port from server instead of picking port from config (issue3746) In order to get the port, wrap create server instead of hgweb_mod/hgwebdir_mod.

branchmap: enable caching for filtered version too The `_branchcache` attribute is turned into a dictionary. Key are filter name and value is a `branchcache` object. Unfiltered version is cached as `None` filter. The attribute is renamed to `_branchcaches` to avoid confusion with the previous one. Both old and new contents are dictionary even if their contents are different. I prefer possible extension code to crash right away instead of just messing the wrong dictionary. As all different caches work isolated to each other, this code keeps the previous behavior of using the unfiltered cache we nothing is filtered. This is a cheap way to have cache collaborate and nullify potential impact in the default case.

branchmap: report filtername when read fails Now that we can have multiple one, we need to know which filecache failed to be read from disk.

branchmap: use a different file name for filtered view of repo

clfilter: ensure unfiltered repo have a filtername attribute too That will allows to use `repo.filtername` for dispatch purpose.

branchmap: move the cache file name into a dedicated function Filtered view of the repo will want to write they file name in a different file.

branchmap: read and write key part related to filtered revision Now that we have a third part for the cache key we need to write and read it on disk. It is only written when there is filtered revision. This keep the format compatible with older version. Notes that, at this state, filtered repository does not use any disk caches yet.

check-code: disallow defunct <> operator Added a test for that and one other python3 check in test-check-code.t.

largefiles: stop using <> operator in favor of != <> has been deprecated for a while, and != is futureproof

commands: fix implicit tuple that is invalid syntax in Python3

check-code: disallow two-argument form of raise Using this old form makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

httpclient: apply change df9aea1def3e: remove use of two-argument raise

test-symlink-os-yes-fs-no.py: clean up use of two-argument raise This makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

scmutil: clean up use of two-argument raise This makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

url: clean up use of two-argument raise This makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

win32: clean up use of two-argument raise This makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

commandserver: clean up use of two-argument raise This makes any attempt to port to Python 3 harder, and the new syntax is supported in 2.4 already.

test-command-template.t: fix test so it all year This test started failing for me after midnight UTC on December 31st. Fixed it by specifying a date 7 years in the future more precisely (rather than just adding 8 to the year and specifying January 1st), which allows the test to pass both now and on 2012-12-01 at the same time.

cmdutil: make getgraphlogrevs limit-aware For a repository with over 400,000 changesets, this speeds up graphlog with a small limit by around 0.05 seconds (~50%).

cmdutil: stop pretending we can calculate revs for graphlog lazily cmdutil.getgraphlogrevs does a ton of work trying to build a graphlog lazily, and then cmdutil.graphlog comes along and destroys all of that. graphmod.dagwalker requires that it be given the full list of revs upfront so that it can perform filtering and tests against known revs. For a repository with over 400,000 changesets, this speeds up graphlog by around 0.02 seconds (~20% with a small limit).

cmdutil: store a local ref to repo.hiddenrevs in getgraphlogrevs On a repository with over 400,000 changesets, this speeds graphlog up by around 0.03 seconds (~20% with a small limit).

cmdutil: make getgraphlogrevs return revs in descending order

branchmap: takes filtered revision in account for cache calculation Tracking tipnode and tiprev is not enough to ensure validaty of the cache as they do not help distinguish a cache that ignored various revisions below tiprev. To detect such difference, we build a hash of all ignored revisions. This hash is then used when checking the validity of a cache for a repo.

branchmap: improve computation of target tip With revision filtering the effective revision number of "tip" may be lower than: len(changelog) - 1 We now use a more correct version preventing useless writing on disk in some case.

branchmap: improve invalid cache message when reading This factors out the generation of the message. This helps future error reporting when reading cache for filtered repository.

histedit: allow operation from non-head if obsolete is enabled Obsolescence markers can represent this situation just fine. Rewritten revisions are marked as precursors of the ones creates by histedit. Unaffected descendants become "unstable". If obsolescence is not enabled we keep the current behavior of aborting. This new behavior only applies when obsolete is enabled and is subject to future discussion and changes.

rebase: allow non-head rebase-set when obsolete is enabled Obsolescence markers can represent this situation just fine. Rebased revisions are marked as precursors of the ones create by rebase. Unrebased descendants becomes "unstable". If obsolescence is not enabled we keep the current behavior of aborting. This new behavior only applies when obsolete is enabled and is subject to future discussion and changes.

amend: allow amend of non-head when obsolete is enabled Obsolescence marker can represent this situation just fine. The old version is marked as precursor of the new changeset. All its descendants become "unstable". If obsolescence is not enabled we keep the current behavior of aborting. This new behavior only applies when obsolete is enabled and is subject to future discussion and changes.

obsolete: factorise troubles detection during push The use of the two methods introduced earlier allows a factorisation of the push code preventing push of troubled changeset.

obsolete: introduce a troubles method on context A troubled changeset may be affected by multiple trouble at the same time. This new method returns a list of all troubles affecting a changes.

obsolete: introduce a troubled method on context Allows to quickly check if a changeset is affected by any troubles. (troubles are: unstable, bumped and divergent)

tests: fix windows test failures

merge with stable

tests: fix windows test failures

largefiles: better test coverage of wireproto and the http protocol

largefiles: cleanup of warnings on errors getting largefiles Especially the "no default or default-push path set in hgrc" was often very misleading and didn't give any hint where it actually was looking. A long error messages is better than several multi-line messages.

largefiles: remove trivial portability wrappers

largefiles: remove reporemove portability wrapper

largefiles: remove findoutgoing portability wrapper

largefiles: use constant for '.hglf/' Problem: '.hglf' + '/' was computed in some tight loops. There is no need for that.

largefiles: use plain wjoin instead of the complex pathto

largefiles: unindent code

largefiles: simplify lfdirstate ignore handling - it is only for tracking .hglf

largefiles: rename 'admin' to more descriptive 'lfstoredir'