revsetbenchmarks: use many more variants by default So far the variants feature was introduced, but not used by default. We now use a set of basic variants by default.

revsetbenchmarks: display even more compact timing result We now use an 8 char display for timing (from 10), we add some logic to drop precision if the number grows too large (as we do not care about sub-0 digit in this case). This allow to pack more variants in a single screen.

revsetbenchmarks: allow running multiple variants per revset The current benchmarks were only testing the whole iteration. This is suboptimal because some changes are meaningful for things like first result, minimum or sorting. We introduce a "variants" feature that let you systematically add some variants to all revsets tested. A typical variants value would be 'plain,min,last,sort'. When testing 'all()' it will also provide testing for: - all() - min(all()) - last(all()) - sort(sort) and output: plain min last sort 0) 0.034568 0.037857 0.000074 0.034238 1) 0.011358 32% 0.020181 53% 0.000080 108% 0.011405 33% Using revsets (who hit the API) instead of the internal API add some overhead, but the overhead should be the same everywhere so it still allow comparison. This is is more simple to implement and allows comparison with older versions who do not have the same API.

revsetbenchmarks: display relative change when meaningful If the time difference is more than 5% from the previous run, we'll display relative information. This makes it much simpler to spot performance changes in a sea of benchmarks.

revsetbenchmarks: improve revision printing We now print the revision number and short hash inline. As a result we drop the crappy list printing.

revsetbenchmarks: hide most timing under a --verbose flag We mostly only care about total time. Dropping this output give us some room to display more useful information (like percentage different) in future changesets.

revsetbenchmarks: drop outdated comment We are no longer testing against system mercurial for quite some time.

revsetbenchmarks: fix argument parsing The file doc was saying something, the code was doing something else, the argument validation was doing a third thing. Doc and behavior now comply with the argument defined in the code.

revsetbenchmarks: use a more compact output format with a header We change the output from: revset #0: draft() 0) wall 0.011989 comb 0.010000 user 0.000000 sys 0.010000 (best of 177) 1) wall 0.012226 comb 0.010000 user 0.000000 sys 0.010000 (best of 193) 2) wall 0.011838 comb 0.020000 user 0.000000 sys 0.020000 (best of 208) to: revset #0: draft() wall comb user sys count 0) 0.012028 0.010000 0.000000 0.010000 170 1) 0.012218 0.010000 0.000000 0.010000 157 2) 0.012622 0.010000 0.000000 0.010000 189 This opens the road to more useful output.

revsetbenchmarks: clarify comment based on irc discussion

revsetbenchmarks: ensure all indexes have the same width This avoids an alignment glitch.

revsetbenchmarks: factor out result output into a function This will make update of the output easier.

revsetbenchmarks: parse perfrevset output into actual number We cannot just ask perfrevset to provide debug output because we usually want to compare output from old version of Mercurial that do not support it. So, we are using a regular expression. (/we now have \d problems/).

revsetbenchmarks: improve error output in case of failure This helps with diagnostics.

revsetbenchmarks: extract call to mercurial into a function This is a gratuitous change to make the code easier to look at.

phases: really fix native phase computation For some reason (probably rebase issue, leprechaun or badly resolved .rej) 1635579f9baf contains only half of the emailed patches and do not fix the bug. This patch adds the other half and enable the sweet native computation for real. As expected this provide massive speedup along the board. revset #0: not public() plain first 0) 0.011960 0.010523 1) 0.000465 3% 0.000492 4% revset #1: (tip~1000::) - public() plain first 0) 0.025700 0.025169 1) 0.002864 11% 0.001899 7% revset #2: not public() and branch("default") plain first 0) 0.022842 0.020863 1) 0.011418 49% 0.010948 52% However, it has a less impact (even bad) on first result time in simple situation. This comes from the overhead of building the set and filtering it. This is especially true on my Mercurial repository (used here) where about 1/3 of the changesets are non public and hidden. This could be mitigated by a caching of the set and a better usage of smartset in '_notpublic'. (But this won't happen in this patch because the win is massive everywhere else). revset #0: not public() last 0) 0.000081 1) 0.000493 x6.1 <-- bad impact revset #1: (tip~1000::) - public() last 0) 0.013966 1) 0.002737 19% revset #2: not public() and branch("default") last 0) 0.011021 1) 0.011038 The effect mostly disappear when the number of non-public changesets is small and/or the repo get bigger. Result for Mozilla central: Mozilla revset #0: not public() plain first last 0) 0.092787 0.084094 0.000080 1) 0.000054 0% 0.000083 0% 0.000083 revset #1: (tip~1000::) - public() plain first last 0) 0.215607 0.183996 0.124962 1) 0.031620 14% 0.006616 3% 0.031168 24% revset #2: not public() and branch("default") plain first last 0) 0.092626 0.082687 0.000162 1) 0.000139 0% 0.000165 0% 0.000167

hgweb: don't point file links at tip hash where it doesn't make sense Some pages, e.g. bookmarks, help and summary don't have a meaningful revision context: they always either show information about tip or about the whole repo (and not about any specific changeset). And error pages can just show hgweb error messages, not related to any repo or changeset. Having a hash in the links worked (even when '{node|short}' resolved to an empty string on error pages), but seeing pages without revision context provide links with hashes is a bit confusing (unless you keep current tip hash in your head at all times) and not consistent with other template styles and other links on the same page: they don't have a hash. Let's just link to '/file', which is equal to '/file/tip'.

hgweb: don't point graph links at tip hash where it doesn't make sense Some pages, e.g. bookmarks, help and summary don't have a meaningful revision context: they always either show information about tip or about the whole repo (and not about any specific changeset). And error pages can just show hgweb error messages, not related to any repo or changeset. When monoblue style was added in 91b0ada2d94b, however, all graph links had tried to point at some hash, and on such pages as described above it didn't make sense. On error pages '{node|short}' is empty string anyway. Of course, it worked, but seeing such pages without revision context provide links with hashes is a bit confusing (unless you keep current tip hash in your head at all times) and wasn't consistent with other template styles, other pages in monoblue and even other links on the same page. Let's just link to '/graph', which is equal to '/graph/tip'.

hgweb: put help link in paper/search.tmpl separately for consistency Just a cosmetic markup change, no .css changes required.

help: use 'color' as an example (instead of 'progress') Progress is now deprecated, using it as an example is suboptimal.

progress: deprecate the progress extension Activating it is a absolute no-op now.

progress: empty the extension of any logic The default value match the one enforce by the extension, we can remove that logic.

progress: move config help into core config help This is core feature now.

progress: display progress bars by default with core Mercurial As discussed multiple time during sprint, we should activate progress by default. Having progress bar significantly improve the user experience.

bundle2: provide number of changesets information to 'addchangegroup' We can now link the two efforts and provided more useful information when pulling changesets.

addchangegroup: accept an expected total number of changesets as argument Caller can optionally informs how much changesets are expected to be added. This will be used for a more useful progress bar output.

getbundle: add data about the number of changesets bundled We use an advisory parameters to carry the number of changesets bundled. This will be used for progress output.

tests: test basic template support for status

templates: add a default template style for status Color doesn't work yet, so no labels here.

formatter: add template support This lets all the non-log commands that use the formatter use templates. There are still some things that don't work, for instance: - color (needs "repo" in map) - shortest (needs "ctx" in map)

formatter: add a method to build a full templater from a -T option

formatter: move most of template option helper to formatter We want to share this function between formatter and cmdutils. It doesn't belong in templater because it imports knowledge of ui layers that shouldn't be there. We'd prefer cmdutil to layer on the formatter rather than vice-versa. Since the formatter is the handler for -T options for all non-log commands, let's move the helper there. We leave the bits specific to the old --style option behind.

color: copy docstring of label() template function to wrapper Otherwise label() wouldn't be listed in "hg help template" if color extension is enabled.

templater: make pad function evaluate both string and rawstring templates "pad" function and "rawstring" type were introduced in parallel, aa51392da507 in default and 5ab28a2e9962 in stable respectively. Therefore, "pad" function lacked handling of "rawstring" unintentionally.

largefiles: ignore hidden changesets with 'verify --large --lfa' Previously, if there were any hidden changesets, the --lfa argument would cause the command to abort with a hint about using --hidden when it tripped over a hidden changeset.

bundle2: clarify in docstring that header size is for a single header The docstring for the header size field currently says "The total number of Bytes used by the part headers", but the size is about a single header, so let's change it to "header".

bundle2: rename duplicate handlepushkeyreply to handleobsmarkerreply The function was only called through the parthandlermapping dict, so it was confusing but safe in practice.

changegroup: remove 'getchangegroupraw' function There is no remaining caller for this function.

exchange: expand usage of getchangegroupraw The 'getchangegroupraw' is very simple (two lines) so we inline it in its only caller. This exposes the 'outgoing' object of the part generator function, allowing us to add information on the number of changesets contained in the part in a later changeset. Such information is useful for progress bar.

getbundle: have a single getchangegroupraw call site Having a single call site will simplify the code and help with coming refactoring.

phases: abort the whole push if phases fail to update (BC) When using bundle2, the phase pushkey parts are now made mandatory. As a result, failure to update the bookmark server side will result in the transaction being aborted.

bookmarks: abort the whole push if bookmarks fails to update (BC) When using bundle2, the bookmark's pushkey parts are now made mandatory. As a result failure to update the bookmark server side will result in the transaction being aborted.

httppeer: allow extensions to replace urllib2.Request The authentication library my extension wants to use requires using a different opener and a different request builder. This change pulls the call to urllib2.Request out so that my extension can replace it just like it can replace urlopener.

progress: move all logic altering the ui object logic in mercurial.ui The ui object can take care of its progress object logic by itself. test-subrepo-recursion is modified because it is a bit sensitive to the "no progress bar" default. It will become unnecessary in the next step when progress will be on by default in core.

progress: move the singleton logic to the ui module The use of a singleton for all of progress handling is debatable (because config may vary). However this is how the extension has been doing it so far. We move that code into the ui module because this is where is should belong when progress is moved into core.

progress: move most extension code into a 'mercurial.progress' module This initiate the relocation of progress into core.

test-subrepo-recursion: restore globs for Windows

tests: restore 'python' and '$TESTDIR/' for dummyssh invocation This is a backout of 46727fea7a00, and a partial backout of c3ecbf694904. Windows won't execute 'dummyssh' directly, presumably because CreateProcess() doesn't know how to execute a bash script: $ hg clone -e "dummyssh" ssh://user@dummy/cloned sshclone remote: 'dummyssh' is not recognized as an internal or external command, remote: operable program or batch file. abort: no suitable response from remote hg! [255] With the restoration of python as the executable, $TESTDIR needs to be restored for these invocations, because python won't search $PATH for 'dummyssh': $ hg clone -e "python dummyssh" ssh://user@dummy/cloned sshclone remote: python: can't open file 'dummyssh': [Errno 2] No such file or directory abort: no suitable response from remote hg! [255]

perf: support -T for every perf commands We are already building a formatter, we can now pass it options the official way.

bundle2: convey PushkeyFailed error over the wire We add a way to convey the precise exception. This will allow better error message on the server.

bundle2: also capture reply capability on failure When unbundling over the wire is aborted, we have a mechanism to convey the error inside a bundle part. As we add support for more errors, we need to know if the client will support them. For this purpose, we duck punch the reply capabilities of the client on the raised extensions. This is similar to what is done to salvage the server output on error.

bundle2: add an 'error' capability This capability will be extended as new error type is introduced.

merge with stable

templater: introduce indent function

hgewb: disable progress when serving (issue4582) Before this patch, progress bar could be displayed when serving, creating hypothetical problems.

test-subrepo-recursion.t: fix progress output on no-hardlink systems

test-ssh: stablize for platform-specific shell quoting Windows and OpenVMS use double quotes instead of single.

push: catch and process PushkeyFailed error We add a way to register "pushkey failure callback" that will be used if the push is aborted by a pushkey failure. A part generator adding mandatory pushkey parts should register a failure callback for all of them. The callback will be in charge of generating a meaningful abort if this part fails. If no callback is registered, the error is propagated. Catch PushkeyFailed error in exchange.

bundle2: introduce a PushkeyFail error to abort unbundle on pushkey error The pushkey code is generic and the server side has little context on what the client is trying to achieve. Generating interesting error messages server side would be challenging. Instead we introduce a dedicated exception that carries more data. In particular, it carries the id of the part which failed that will allow clients to display custom error messages depending on the part intent. The processing and transfer-over-the-wire of this exception is to be implemented in coming changesets.

record: exiting editor with non-zero status should not stop recording session Before this patch, exiting a hunk edit in record with a non-zero status lead to the end of the recording session, losing previously-selected hunks to record. This patch introduces the more desirable behavior of warning the user and continuing the recording session.

progress: stop double-wrapping of ui class We were wrapping the ui class again and again (uisetup, reposetup, subrepo setup, remote repo setup, etc). We now avoid that. This has impact on tests that were double-printing data because of this.

bundle2: abort when a mandatory pushkey part fails So far, result of a pushkey operation had no consequence on the transaction (beside the change). We makes it respect the 'mandatory' flag of part so that failed pushkey call abort the whole transaction. This will allow rejecting changes (primary target: changesets) regarding phases or bookmark criteria in the future (when we will push such data in a mandatory part). We currently raise an abort error because all clients support it. We'll introduce a more precise error in the next changesets.

push: make pushkey part advisory The current behavior (with bundle1) is to let the rest of the push succeed if the pushkey call (phases, bookmarks) failed (this comes from the fact that each item is sent in its own command). We kept this behavior with bundle2, which is highly debatable, but let us keep thing as they are now as a start. We are about to enforce 'mandatory' pushkey part as 'mandatory' successful, so we need to marks parts as advisory to preserve the current (debatable) behavior.

bundle2: pull bookmark the old way if no bundle2 listkeys support (issue4701) All known server implementations have listkeys support with bundle2, but people in the process of implementing new servers may not. Let's be nice with them.

tests: simplify printenv calls Make printenv executable so that we don't need python, TESTDIR, or quoting.

tests: make printenv executable

tests: simplify calls to dummyssh dummyssh is marked executable and is in the path, no need for python, TESTDIR, or quotes.

tests: remove duplicate inline dummyssh in test-wireproto.t

tests: drop DAEMON_PIDS from killdaemons calls

tests: make killdaemons.py use DAEMON_PIDS by default

tests: drop explicit $TESTDIR from executables $TESTDIR is added to the path, so this is superfluous. Also, inconsistent use of quotes means we might have broken on tests with paths containing spaces.

templater: do not preprocess template string in "if" expression (issue4714) The problem was spotted at 5ab28a2e9962, that says "this patch invokes it with "strtoken='rawstring'" in "_evalifliteral()", because "t" is the result of "arg" evaluation and it should be "string-escape"-ed if "arg" is "string" expression." This workaround is no longer valid since 890845af1ac2 introduced strict parsing of '\{'. Instead, we should interpret bare token as "string" or "rawstring" template. This is what buildmap() does at parsing phase.

largefiles: use the optional badfn argument when building a matcher The monkey patching in cat() can't be fixed, because it still delegates to the original bad(). Overriding commands.cat() should go away in favor overriding cmdutil.cat() anyway, and that matcher can be wrapped with matchmod.badmatch().

mq: use the optional badfn argument when building a matcher

commands: use the optional badfn argument when building a matcher

scmutil: add an optional parameter to matcher factories for a bad() override Even though scmutil.matchandpats() is documented to warn about bad files, several callers silence the warning.

scmutil: use the optional badfn argument when building a matcher

context: add an optional constructor parameter for a match.bad() override Most matcher creation is done by way of a context.

match: add an optional constructor parameter for a bad() override This will be used to eliminate monkey patching of new matcher instances that weren't removed in 5984dd42e140::1a95c57959f6.

mail: pass ui to sslutil.wrapsocket() even if verifycert is off (issue4713) 21b536f01eda made 'ui' argument is passed via sslutil.sslkwargs(), but mailer doesn't call sslkwargs() if smtp.verifycert is off. So we have to put it in sslkwargs manually.

windows: use os.SEEK_END

merge with stable

crecord: fix three typos introduced while moving crecord into core When moving crecord to core, I did a search an replace to remove all of the variable with caps letter. Doing so, I unfortunately changed some commands with capital letter shortcut to their lowercase counterpart. I fixed one of these cases in 093d38165e5a. This patch fixes all of the remaining cases.

revlog: move size limit check to addrevision This lets us add the name of the indexfile to the message.

sshpeer: also use doublepipe for client to server communication This will allow even more real time output when the server issue output in the middle a stream push.

sshpeer: allow doublepipe on unbuffered main pipe The output pipe does not have manually managed read buffer (actually, no read anything). To also use the doublepipe for outgoing write (useful to consume remote output when pushing large set of data) we need the doublepipe to work on standard pipe too.

sshpeer: allow write operations through double pipe We have a shiny toy, lets make it wider.

sshpeer: rename 'size' to 'data' in doublepipe We are about to add 'write' support, the argument will be either an int or a string.

mq: ban \r and \n in patch names (issue4711) This is at best crazy, and at worst will break things like the series file. Let's just stop the madness.

mq: use %r to format illegal characters instead of manually quoting This will make it easier to ban \r and \n in the next patch and still have a sensible error message.

histedit: abort rather than edit a public changeset (issue4704) This is suboptimal as the user still has to explicitly cancel the histedit afterwards, but it prevents the immediate problem. histedit should probably implicitly do 'hg histedit --abort' if a util.Abort is raised internally.

test-histedit-edit: add test that demonstrates bug in issue4704 Durham and I agree that it's a bug you can fold into a change that's not listed in the histedited set, so we'll follow this up with a change that prevents folds as the first edit step.

histedit: copyedit docstring wording problem I noticed while here

merge with stable

dirstate: avoid invalidating every entries when list is empty Default value was not tested with 'is None', this made empty list seen as default value and result the invalidation of every single entry in the dirstate. On repos with hundred of thousand of files, this results in minutes of lookup time instead nothing. This is a text book example of why we should test 'is None' if this is what we mean.

crecord: fix a typo introduced when moving crecord to core When moving crecord to core, I did a search an replace to remove all of the variable with caps letter. Doing so, I unfortunately changed the 'F' command to 'f' leading to two 'f' commands, one never used. This patch fixes this mistake and re enables the behavior of 'F'.

pull: prevent race condition in bookmark update when using -B (issue4689) We are already fetching remote bookmarks to honor the -B option, we now pass that data to the pull process so it can reuse it. This prevents a race condition between the initial looking and the actual pulling of changesets and bookmarks. Tests are updated to handle this fact.

pull: allow a generic way to pass parameters to the pull operation We have been feeling the need for this in extensions for quite some time. This will be used to pass remote bookmark information around in the next changesets.

pull: skip pulling remote bookmarks with bundle2 if a value already exists For efficiency and consistency purpose, remote bookmarks, retrieved at the time the pull command code is doing lookup, will be reused during the core pull operation. A second step toward this is to avoid requesting bookmark information in the bundle 2 if we already have them locally.

pull: skip pulling remote bookmarks with bundle1 if a value already exist For efficiency and consistency purpose, remote bookmarks, retrieved at the time the pull command code is doing lookup, will be reused during the core pull operation. A first step toward this is to setup the logic avoiding pulling the data again during the discovery phase if some have already been provided.

test: pull through http when testing for race conditions The http server is stateless giving more occasion for race. We switch the test to http before adding extra cases tests.

test: display pre-pull remote bookmark state when testing race condition This makes it easier to read the test. This will be usefull for comming changesets testing more race conditions.

largefiles: replace match.bad() monkey patching with match.badmatch() No known issues with the previous code since it restored the original method, but this is cleaner. The monkey patching in cat is harmless, because it is created locally, and doesn't pass it anywhere (subrepo cat isn't supported with largefiles).

revert: replace match.bad() monkey patching with match.badmatch() No known issues with the previous code since it immediately overwrote the patched, locally create matcher.

cat: replace match.bad() monkey patching with match.badmatch() No known issues with the previous code since it restored the original method, but this is cleaner.

forget: replace match.bad() monkey patching with match.badmatch() The previous code didn't restore the original method, but it looks like the worst that would happen is junk added to a list that had already been processed by previous subrepo invocation(s).

add: replace match.bad() monkey patching with match.badmatch() The previous code didn't restore the original method, but it looks like the worst that would happen is junk added to a list that had already been processed by previous subrepo invocation(s).

context: replace match.bad() monkey patching with match.badmatch() No known issues with the previous code since it restored the original method, but this is cleaner.

addremove: replace match.bad() monkey patching with match.badmatch() No known issues with the previous code since it restored the original method, but this is cleaner.

match: introduce badmatch() to eliminate long callback chains with subrepos Various bit of code replace the bad method on matchers, and then delegate to the original bad method after doing some custom processing. At least some of these forget to restore the original method when the need has passed, and then when the matcher is passed to the next subrepo (even a sibling), another layer is added such that the chain looks like: bad2 -> bad1 -> original At best, this is a waste of processing, but sometimes spurious messages can be emitted (e.g. ccb1623266eb). The trick with this copy of the matcher is to make sure it is *not* passed to any subrepo- the original must be passed instead.

ssl: remove CERT_REQUIRED constant that was necessary for compatibility

ssl: drop try-except clause that was necessary for ancient Python

ssl: drop support for Python < 2.6, require ssl module try-except clause is kept for readability of this patch, and it will be removed soon.

ssl: rename ssl_wrap_socket() to conform to our naming convention I've removed ssl_ prefix because the module name contains ssl.

test-https: kill only hgweb daemon to restart for client-auth test This avoids a test failure on Mac OS X. Because tinyproxy.py isn't detached from the shell, it may complain that a child process is terminated by a signal.

merge with stable

hgwebdir: don't allow the hidden parent of a subrepo to show as a directory Previously, if a subrepo parent had 'web.hidden=True' set, neither the parent nor child had a repository entry. However, the directory entry for the parent would be listed (it wouldn't have the fancy 'web.name' if configured), and that link went to the repo's summary page, effectively making it not hidden. This simply disables the directory processing if a valid repository is present. Whether or not the subrepo should be hidden is debatable, but this leaves that behavior unchanged (i.e. it stays hidden).

pull: avoid race condition with 'hg pull --rev name --update' (issue4706) The previous scheme was: 1) lookup node for all pulled revision, 2) pull said node 3) lookup the node of the checkout target 4) update the repository there. If the remote repo changes between (1) and (3), the resolved name will be different and (3) crash. There is actually no need for a remote lookup during (3), we could just set the value in (1). This prevent the race condition and save a possible network roundtrip.

revert: add an experimental config to use inverted selection We had a discussion on the list about the interactive ui for revert. This patch adds a flag to allow people to test the second alternative (referred to as proposition 2 on the mailing list). It effectively inverts the signs in the

bundle2: stop capturing output for ssh again This backout 36111f98f23d since we can have real time output with ssh again. The tests change is not backed-out because it was a test output fix.

sshpeer: use the doublepipe object for the server to client channel This restores real-time output from ssh server while waiting for protocol data sent by the server.

sshpeer: introduce a "doublepipe" class This class is responsible for ensuring we still process the server output streamed through the ssh's 'stderr' pipe during the initial wait for other protocol streams. It currently only works on posix system because of its use of 'select.select'.

util: add a simple poll utility We'll use it to detect when a sshpeer have server output to be displayed. The implementation is super basic because all case support is not the focus of this series.

crecord: fix blue artifact bug coming back from help screen Before this patch, with record's curses interface: - When one pressed the ? key to see the help menu and then leaves the help menu - Then, blue artifacts would remain on the screen This patch forces a screen clear / refresh after coming back from the help menu. Since we don't have UI test for crecord I checked on my machine that the bug does not reproduce.

scmutil: consistently return subrepos relative to ctx1 from itersubrepos() Previously, if a subrepo was added in ctx2 and then compared to another without it (ctx1), the subrepo for ctx2 was returned amongst all of the ctx1 based subrepos, since no subrepo exists in ctx1 to replace it in the 'subpaths' dict. The two callers of this, basectx.status() and cmdutil.diffordiffstat(), both compare the yielded subrepo against ctx2, and thus saw no changes when ctx2's subrepo was returned. The tests here previously didn't mention 's/a' for the 'p1()' case. This appears to have been a known issue, because some diffordiffstat() comments mention that the subpath disappeared, and "the best we can do is ignore it". I originally ran into the issue with some custom convert code to flatten a tree of subrepos causing hg.putcommit() to abort, but this new behavior seems like the correct status and diff behavior regardless. (The abort in convert isn't something users will see, because convert doesn't currently support subrepos in the official repo.)

context: introduce the nullsub() method Ultimately, this will be used by scmutil. The subrepo module already imports it, so it can't import the subrepo module to access the underlying method.

subrepo: introduce the nullsubrepo() method This will be used in an upcoming patch. It's a one-off use, but seems better to be contained in the subrepo module, than for the next patch to overwrite the _ctx and _state fields in another module. '' is used as the default revision in subrepo.state() if it can't be found, so it seems like a safe choice.

ssl: prompt passphrase of client key file via ui.getpass() (issue4648) This is necessary to communicate with third-party tools through command-server channel. This requires SSLContext backported to Python 2.7.9+. It doesn't look nice to pass ui by sslkwargs, but I think it is the only way to do without touching various client codes including httpclient (aka http2). ui is mandatory if certfile is specified, so it has no default value. BTW, test-check-commit-hg.t complains that ssl_wrap_socket() has foo_bar naming. Should I bulk-replace it to sslwrapsocket() ?

https: do not inherit httplib.HTTPSConnection that creates unused SSLContext HTTPSConnection of Python 2.7.9 creates SSLContext in __init__, which involves a password prompt for decrypting the private key. This means the password was asked twice, one for unused SSLContext, and next for our ssl function. Because our httpsconnection replaces connect() method at all, we can simply drop httplib.HTTPSConnection. Instead, class and instance attributes are copied from it. HTTPSConnection of Python 2.7.8 and 2.6.9 seem to have no such problem. https://hg.python.org/cpython/file/v2.7.9/Lib/httplib.py#l1183

test-https: test basic functions of client certificate authentication Because hgweb doesn't support client certificates, I just patched it to require client certificates that are signed and verified by the server certificate. This won't be ideal for production servers, but should be okay for the test. The encrypted key file will be used by future patches. I couldn't figure out a way to redirect a password prompt provided by OpenSSL, so it isn't tested for now.

phases: add `hg help phases` hint to failures to edit public commits There were a couple of locations that were missing this hint after an edition of some kind failed due to the public phase.

phases: rewrite "immutable changeset" to "public changeset" The phrase "cannot edit immutable changeset" is kind of tautological. Of course unchangeable things can't be changed. We instead mention "public" and provide a hint so that we can point to the actual problem. Even in cases where some operation other than edition cannot be performed, "public" gives the root cause that results in the "immutable" effect. There is a precedent for saying "public" instead of "immutable", for example, in `hg commit --amend`.

revlog: raise an exception earlier if an entry is too large (issue4675) Before we were relying on _pack to error out when trying to pass an integer that was too large for the "i" format specifier. Now we check this earlier so we can form a better error message. The error message unfortunately must exclude the filename at this level of the call stack. The problem is that this name is not available here, and the error can be triggered by a large manifest or by a large file itself. Although perhaps we could provide the name of a revlog index file (from the revlog object, instead of the revlogio object), this seems like too much leakage of internal data structures. It's not ideal already that an error message even mentions revlogs, but this does seem unavoidable here.

wix: move library.zip and all *.pyd into a lib/ folder This makes the root install folder (on Windows) nice and tidy. The only files left in the root folder are: hg.exe python27.dll COPYING.rtf ReadMe.html the last of which was probably out-of-date 7 years ago

sshpeer: run the ssh command unbuffered This is necessary to use non-blocking IO base on polling. Such polling is needed to restore real time output with ssh peer. Changeset fce065538bcf is talking about 5x regression on Mac OS X when playing with this value. So we introduced our own buffering layer in previous changesets. This seems to keep the regression away (we are even issuing much less read).

sshpeer: use a 'bufferedinputpipe' for standard output of the ssh process We need this pipe to still be buffered when will switch to unbuffered pipe. (switch motivated by the need of using polling to restore real time output from ssh server). This is the only pipe that needs to be wrapped because this is the one who do extensive usage of 'readline'. The stderr pipe of the process is alway read in non blocking raw chunk, so it won't benefit from the buffering.

util: introduce a bufferedinputpipe utility To restore real time server output through ssh, we need to using polling feature (like select) on the pipes used to communicate with the ssh client. However we cannot use select alongside python level buffering of these pipe (because we need to know if the buffer is non-empty before calling select). However, unbuffered performance are terrible, presumably because the 'readline' call is issuing 'read(1)' call until it find a '\n'. To work around that we introduces our own overlay that do buffering by hand, exposing the state of the buffer to the outside world. The usage of polling IO will be introduced later in the 'sshpeer' module. All its logic will be very specific to the way mercurial communicate over ssh and does not belong to the generic 'util' module.

test: drop all the forced use of bundle2 Using bundle2 during exchange is now the default, we do not need all this explicit enabling of bundle2 exchange in test anymore.

bundle2: use bundle2 by default All the test change have been isolated and validated. We have free to turn on bundle2 as the default exchange protocol. "To reach a port we must set sail – Sail, not tie at anchor Sail, not drift."

wireprotocol: distinguish list and set in getbundle argument The 'bundlecaps' argument is expected to be a set, but 'listkeys' is expected to be a list where ordering matters. We introduce a new 'scsv' argument type for the 'set' version and move 'csv' to the 'list' version. 'test-ssh.t' is changed because this introduced an instability in the order we were producing listkeys parts.

exchange: support transferring .hgtags fnodes mapping On Mozilla's mozilla-beta repository .hgtags fnodes resolution takes ~18s from a clean cache on my machine. This means that the first time a user runs `hg tags`, `hg log`, or any other command that displays or accesses tags data, a ~18s pause will occur. There is no output during this pause. This results in a poor user experience and perception that Mercurial is slow. The .hgtags changeset to filenode mapping is deterministic. This patch takes advantage of that property by implementing support for transferring .hgtags filenodes mappings in a dedicated bundle2 part. When a client advertising support for the "hgtagsfnodes" capability requests a bundle, a mapping of changesets to .hgtags filenodes will be sent to the client. Only mappings of head changesets included in the bundle will be sent. The transfer of this mapping effectively eliminates one time tags cache related pauses after initial clone. The mappings are sent as binary data. So, 40 bytes per pair of SHA-1s. On the aforementioned mozilla-beta repository, 659 * 40 = 26,360 raw bytes of mappings are sent over the wire (in addition to the bundle part headers). Assuming 18s to populate the cache, we only need to transfer this extra data faster than 1.5 KB/s for overall clone + tags cache population time to be shorter. Put into perspective, the mozilla-beta repository is ~1 GB in size. So, this additional data constitutes <0.01% of the cloned data. The marginal overhead for a multi-second performance win on clones in my opinion justifies an on-by-default behavior.

bundle2: part handler for processing .hgtags fnodes mappings .hgtags fnodes cache entries can be expensive to compute, especially if there are hundreds of even thousands of them. This patch implements support for receiving a bundle2 part that contains a mapping of changeset to .hgtags fnodes. An upcoming patch will teach the server to send this part, allowing clients to bypass having to redundantly compute these values. A number of tests changed due to the client advertising the "hgtagsfnodes" capability.

changegroup: rename _computeoutgoing to computeoutgoing We're going to use this function from another module in an upcoming patch. Drop the _ prefix to mark it as non-private.

merge with stable

test-treemanifest: add globs for Windows

test-import: sync with 6084926366b9 for platforms without execbit

hgwebdir: avoid redundant repo and directory entries when 'web.name' is set Previously, when 'web.name' was set on a subrepo parent and 'web.collapse=True', the parent repo would show in the list with the configured 'web.name', and a directory with the parent repo's filesystem name (with a trailing slash) would also appear. The subrepo(s) would unexpectedly be excluded from the list of repositories. Clicking the directory entry would go right to the repo page. Now both the parent and the subrepos show up, without the additional directory entry. The configured hgweb paths used '**' for finding the repos in this scenario. A couple of notes about the tests: - The area where the subrepo was added has a comment that it tests subrepos, though none previously existed there. One now does. - The 'web.descend' option is required for collapse to work. I'm not sure what the previous expectations were for the test. Nothing changed with it set, prior to adding the code in this patch. It is however required for this test. - The only output changes are for the hyperlinks, obviously because of the 'web.name' parameter. - Without this code change, there would be an additional diff: --- /usr/local/mercurial/tests/test-hgwebdir.t +++ /usr/local/mercurial/tests/test-hgwebdir.t.err @@ -951,7 +951,7 @@ /rcoll/notrepo/e/ /rcoll/notrepo/e/e2/ /rcoll/notrepo/f/ - /rcoll/notrepo/f/f2/ + /rcoll/notrepo/f/ Test repositories inside intermediate directories I'm not sure why the fancy name doesn't come out, but it is enough to demonstrate that the parent is not listed redundantly, and the subrepo isn't skipped.

Added signature for changeset ed18f4acf435

Added tag 3.4.1 for changeset ed18f4acf435

merge with stable

templatekw: compare target context and its parent exactly (issue4690) Before this patch, template keywords `{file_mods}`, `{file_adds}` and `{file_dels}` use values gotten by `repo.status(ctx.p1().node(), ctx.node())`. But this doesn't work as expected if `ctx` is `memctx` or `workingcommitctx`. Typical case of templating with these contexts is customization of the text shown in the commit message editor by `[committemplate]` configuration. In this case, `ctx.node()` returns None and it causes comparison between `ctx.p1()` and `workingctx`. `workingctx` lists up all changed files in the working directory even at selective committing. BTW, `{files}` uses `ctx.files()` and it works as expected. To compare target context and its parent exactly, this patch passes `ctx.p1()` and `ctx` without `node()`-nize. This avoids unexpected comparison with `workingctx`. This patch uses a little redundant template configurations in `test-commit.t`, but they are needed to avoid regression around problems fixed by a4958cdb2202 and 1e6fb8db666e: accessing on `ctx` may break `ctx._status` field.

test: use bundle2 in test-pull-http This test has a minor output change when using bundle2. We make this change early to reduce the noise in the final change.

test: use bundle2 in test-push-http Now that we have a bundle1 version of this test, we can move the original to bundle2. This lets us handle the ouput change from using the bundle2 protocol earlier.

test: copy test-push-http.t to testpush-http-bundle1.t We want to keep both code paths tested. The test is a bit too extensive to simply introduce dual testing in it, so we make a copy for each protocol version.

tests: add (?) flag for optional lines When the test engine fails to match output on a line marked with (?), it will simply continue to the next expected line and try again. This allows simplifying tests that have either version-specific or non-fixed behavior, for instance: $ coin-flip heads (?) tails (?) (There's no form of back-tracking attempted, so optional matches should be specific.)

templatekw: display active bookmark more consistently (issue4552) (BC) Previously, the template keyword '{activebookmark}' would only display the active bookmark if it was also pointing to the working directory's parent. Meanwhile, the '{active}' subkeyword of the '{bookmarks}' keyword displays the active bookmark regardless of whether it also points to the working directory's parent. This is confusing. Consider the output of these two templates: $ hg log -T '{activebookmark}\n' -r indent $ hg log -T '{bookmarks % "{bookmark}"}\n' -r indent indent This is the current behavior that can arise after, eg, a pull moves a bookmark out from under you. After this patch, the first template will also return the active bookmark that points to a revision, even if it is not the current parent of the working directory. A test has been added to show the new behavior.

revrange: build spanset from x:y range This slightly improves the performance in the optimal case: % hg log -R mozilla-central -r0:tip -l1 --time (before) time: real 0.050 secs (user 0.040+0.000 sys 0.010+0.000) (after) time: real 0.020 secs (user 0.000+0.000 sys 0.010+0.000)

revrange: build balanced tree of addsets from revisions (issue4565) This reduces the stack depth from O(n) to O(log(n)). Therefore, repeated -rREV options will never exceed the Python stack limit. Currently it depends on private revset._combinesets() function. But at some point, we'll be able to drop the old-style parser, and revrange() can be completely rewritten without using _combinesets(): trees = [parse(s) for s in revs] optimize(('or',) + trees) # combine trees and optimize at once ... Blockers that prevent eliminating old-style parser: - nullary ":" operator - revrange(repo, [intrev, ...]), can be mapped to 'rev(%d)' ? - revrange(repo, [binnode, ...]), should be banned ?

revrange: clean up meaningless reconstruction of sets They just exist for deduplication that was removed by the previous patch.

revrange: drop unnecessary deduplication of revisions Because "l" is a smartset, duplicated entries are omitted by addsets.

summary: move the parents phase marker to commit line (issue4688) The phase of the pending commit depends on the parent of the working directory and on the phases.newcommit configuration. First, this information rather depend on the commit line which describe the pending commit. Then, we only want to be advertised when the pending phase is going to be higher than the default new commit phase. So the format will change from $ hg summary parent: 2:ab91dfabc5ad foo parent: 3:24f1031ad244 tip bar branch: default commit: 1 modified, 1 unknown, 1 unresolved (merge) update: (current) phases: 1 secret (secret) to parent: 2:ab91dfabc5ad foo parent: 3:24f1031ad244 tip bar branch: default commit: 1 modified, 1 unknown, 1 unresolved (merge) (secret) update: (current) phases: 1 secret

tags: support setting hgtags fnodes cache entries An upcoming patch will teach bundle2 to transfer .hgtags fnodes values. To support this, we need to support inserting values into the cache. Add functionality to do that.

tags: support reading tags cache without populating An upcoming patch will teach the bundle2 protocol to transfer .hgtags fnodes to the client. We don't want this to incur any extra work at serve time. Create an optional cache query mode that doesn't populate the cache as a side-effect.

check-commit: make foo_bar naming regexp less greedy \s is equivalent to the character class [ \t\n\r\f\v]. Using \s+ in a regular expression against input with multiple lines may match across multiple lines. For the regexp in question, "\+\s+" would match "+\n " and similar sequences, leading to false positives for functions that were included in diff context, after a modified hunk.

test: use bundle2 in test-http Now that we have a bundle1 version of this test, we can move the original to bundle2. This lets us handle the ouput change from using the bundle2 protocol earlier.

test: copy test-http.t to test-http-bundle1.t We want to keep both code paths tested. The test is a bit too extensive to simply introduce dual testing in it so we make a copy for each protocol version.

test: use bundle2 in test-ssh Now that we have a bundle1 version of this test, we can move the main version to bundle2. This lets us handle the ouput change from using the bundle2 protocol earlier.

test: copy test-ssh.t to test-ssh-bundle1.t We want to keep both code paths tested. The test is a bit too extensive to simply introduce dual testing in it so we make a copy for each protocol version.

test: lock test-unbundlehash to bundle1 usage It is testing a bundle1 specific behavior. Bundle2 has its own way there. See inline comment for details.

test: use bundle2 in test-acl This test makes extensive use of --debug so moving to bundle2 based exchange has a massive impact. We do it early to reduce the noise create by a future usage of bundle2 as the default protocol.

test: use both bundle formats in test-pull-http It is valuable to have both formats tested.

test: use bundle2 in test-http-proxy The proxy test does not care about what protocol is used, but the new protocol implies different traffic (and therefore different log output). We switch it to bundle2 early to minimise the noise of using bundle2 for exchange by default.

tests: use bundle2 for test-hook Using bundle2 has an effect on which hooks are run when. We turn it on for test-hooks early to reduce the noise of switching the default exchange to bundle2.

pull: only prefetch bookmarks when using bundle1 All bundle2 servers now support the 'listkeys' part(1), so we'll always be able to fetch bookmarks data at the same time as the changeset. This should be enough to avoid the one race condition that this bookmark prefetching is trying to work around. It even allows future server to make sure everything is generated from the same "transaction" if they become capable of such. The current code was already overwriting the prefetched value with the one in bundle2 anyway. Note that this is not preventing all race conditions in related to bookmark in 'hg pull' it makes nothing better and nothing worse. Reducing the number of listkeys calls will reduce the latency on pull. The pre-fetch is also moved into a discovery step because it seems to belong there. (1) Because all servers not speaking 'pushkey' parts are compatible with the 'HG2X' protocol only.

pull: document the race condition with bookmark name It seems valuable to document this in-place to help the next poor soul looking at this code to understand what kind of beast he is currently facing.

pull: only list remote bookmarks if -B is used to populate pulled heads Listing remote bookmarks results in network traffic and latency. This should be avoided when possible.

i18n-ja: synchronized with 8594d0b3018e

hg: explicitly check that peer lookup object has instance() if call failed If a "thing" is callable but raises TypeError for some reason, a callable object would be returned. Thereafter, unfriendly traceback would be displayed: Traceback (most recent call last): ... File "mercurial/hg.pyc", line 119, in _peerorrepo obj = _peerlookup(path).instance(ui, path, create) AttributeError: 'function' object has no attribute 'instance' Instead, we should show the reason why "thing(path)" didn't work: Traceback (most recent call last): ... File "hggit/__init__.py", line 89, in _local p = urlcls(path).localpath() TypeError: 'NoneType' object is not callable If a "thing" is not callable, it must be a module or an object that implements instance(). If that module didn't have instance(), the error message would be "<unloaded module 'foo'> object is not callable". It doesn't make perfect sense, but it isn't so bad as it can blame which module went wrong.

extensions: show traceback on load failure if --traceback flag is set Before this patch, there was no handy way to investigate the reason why extension couldn't be loaded. If ui.debug is set, tracebacks of both "hgext.foo" and "foo" are displayed because the first ImportError could occur at very deep dependency module.

ui: flush stdout before writing traceback to stderr Sometimes a traceback message is paired with ui.debug(). This patch makes sure that these messages are displayed in the right order.

treemanifest: visit directory 'foo' when given e.g. '-X foo/ba?' For globs like 'foo/ba?', match._roots() will return 'foo'. Since visitdir(), excludes directories in the excluded roots, it would skip the entire foo directory. This is incorrect, since 'foo/ba?' doesn't mean that everything in foo/ should be exluded. Note that visitdir() is called only from the treemanifest class, so this only affects tree manifests. Fix by adding roots to the set of excluded roots only if there are no excluded patterns. Since 'glob' is the default pattern type for globs, we also need to update some -X patterns in the tests to be of 'path' type to take advantage of the visitdir tricks. For consistency, also update the -I patterns. It seems a little unfortunate that 'foo' in 'hg files -X foo' is considered a pattern because of the implied 'glob' type, but improving that is left for another day.

phases: fix bug where native phase computation wasn't called I forgot to include this change as a previous diff and the native code to compute the phases was never called. The AttributeError was silently caught and the pure implementation was used instead.

merge with stable

record: add default value for operation argument This patch is part of a series of patches to change the recording ui to reflect the operation currently running (commit, shelve, revert ...). This patch adds the default value of the operation argument for record's standard and curses interface to match what is displayed in the interface as of today.

record: precise documentation This patch improves the documentation of the recordfilter function to explain that we need a translated string for the 'operation' argument.

test: use both bundle formats in test-bookmarks-pushpull It is valuable to have both formats tested.

test: explicitly test wire protocol discovery with bundle1 and bundle2 We want to make sure the behavior is similar in both cases.

test: use bundle2 for exchange in test-push-warn Using bundle2 for exchange unlocks the usage of changegroup version 2. This version of the changegroup held more information (delta base) that result in a small increase in content size (20 bytes).

test: use bundle2 for exchange in test-largefile Using bundle2 for exchange unlock the usage of changegroup version 2. This version of the changegroup held more information (delta base) that result in a small increase in content size. It also lead to reordering of output because the exchange scheduling is a bit different.

test: use bundle2 for exchange in test-push-warn Using bundle2 for exchange unlocks the usage of changegroup version 2. This version of the changegroup held more information (delta base) that result in a small increase in content size (20 bytes x 2 changes).

test: use bundle2 for exchange in test-bookmark-pushpull This lets us adapt to changes in obsmarkers exchange output earlier.

test: use bundle2 for exchange in test-obsolete The bundle2 version of obsmarkers exchange is more informative. Switching to bundle2 by default will change the output of this tests. To reduce the noise when switching bundle2 to the default protocol, we migrate this tests early.

revert: remove unused debug code in the test file I left a test environment variable in a previous commit and it is doing nothing useful. This patch removes it.

commands: rename current to active in variables and comments Today, the terms 'active' and 'current' are interchangeably used throughout the codebase in reference to the active bookmark (the bookmark that will be updated with the next commit). This leads to confusion among developers and users. This patch is part of a series to standardize the usage to 'active' throughout the mercurial codebase and user interface.

templatekw: introduce active subkeyword from bookmarks keyword Today, the terms 'active' and 'current' are interchangeably used throughout the codebase in reference to the active bookmark (the bookmark that will be updated with the next commit). This leads to confusion among developers and users. This patch is part of a series to standardize the usage to 'active' throughout the mercurial codebase and user interface.

bookmarks: name label for active bookmark correctly Retain old label as well for backwards compatibility. Today, the terms 'active' and 'current' are interchangeably used throughout the codebase in reference to the active bookmark (the bookmark that will be updated with the next commit). This leads to confusion among developers and users. This patch is part of a series to standardize the usage to 'active' throughout the mercurial codebase and user interface.

tests: descending empty dirs works in all hgweb styles, test them too The tested feature was added to multiple hgweb styles in c21d236ca897, but only paper was tested. Let's test everything now, including monoblue, which only got the feature some 6 years late in e50d8b21f4f4.

tests: actualize the comment in test-hgweb-descend-empties.t The comment came together with the whole test file and the feature (descend empty dirs in hgweb) in c21d236ca897, but for some reason wasn't exactly accurate. Namely, there isn't e1 directory in the test at all, it obviously should say d1; and b1 didn't terminate at level 3, but does now.

revset: add fast path for _list() of integer revisions This can greatly speed up chained 'or' of integer revisions. 1) reduce nesting of chained 'or' operations 2) optimize to a list 3) fast path for integer revisions (this patch) revset #0: 0 + 1 + 2 + ... + 1000 1) wall 0.483341 comb 0.480000 user 0.480000 sys 0.000000 (best of 20) 2) wall 0.025393 comb 0.020000 user 0.020000 sys 0.000000 (best of 107) 3) wall 0.008371 comb 0.000000 user 0.000000 sys 0.000000 (best of 317) revset #1: sort(0 + 1 + 2 + ... + 1000) 1) wall 0.035240 comb 0.040000 user 0.040000 sys 0.000000 (best of 100) 2) wall 0.026432 comb 0.030000 user 0.030000 sys 0.000000 (best of 102) 3) wall 0.008418 comb 0.000000 user 0.000000 sys 0.000000 (best of 322) revset #2: first(0 + 1 + 2 + ... + 1000) 1) wall 0.028949 comb 0.030000 user 0.030000 sys 0.000000 (best of 100) 2) wall 0.025503 comb 0.030000 user 0.030000 sys 0.000000 (best of 106) 3) wall 0.008423 comb 0.010000 user 0.010000 sys 0.000000 (best of 319) But I admit that it is still slower than the spanset. revset #3: 0:1000 3) wall 0.000132 comb 0.000000 user 0.000000 sys 0.000000 (best of 19010)

revset: optimize 'or' operation of trivial revisions to a list As seen in issue4565 and issue4624, GUI wrappers and automated scripts are likely to generate a long query that just has numeric revisions joined by 'or'. One reason why is that they allows users to choose arbitrary revisions from a list. Because this use case isn't handled well by smartset, let's optimize it to a plain old list. Benchmarks: 1) reduce nesting of chained 'or' operations 2) optimize to a list (this patch) revset #0: 0 + 1 + 2 + ... + 1000 1) wall 0.483341 comb 0.480000 user 0.480000 sys 0.000000 (best of 20) 2) wall 0.025393 comb 0.020000 user 0.020000 sys 0.000000 (best of 107) revset #1: sort(0 + 1 + 2 + ... + 1000) 1) wall 0.035240 comb 0.040000 user 0.040000 sys 0.000000 (best of 100) 2) wall 0.026432 comb 0.030000 user 0.030000 sys 0.000000 (best of 102) revset #2: first(0 + 1 + 2 + ... + 1000) 1) wall 0.028949 comb 0.030000 user 0.030000 sys 0.000000 (best of 100) 2) wall 0.025503 comb 0.030000 user 0.030000 sys 0.000000 (best of 106)

revset: make "null" able to appear in internal _list() expression This is the same workaround introduced at e16456831516. Without this patch, "null or x" can't be optimized to _list(null x). Test case will be added by the next patch.

revset: make internal _list() expression remove duplicated revisions This allows us to optimize chained 'or' operations to _list() expression. Unlike _intlist() or _hexlist(), it's difficult to remove duplicates by the caller of _list() because different symbols can point to the same revision. If the caller knows all symbols are unique, that probably means revisions or nodes are known, therefore, _intlist() or _hexlist() should be used instead. So, it makes sense to check duplicates by _list() function. '%ls' is no longer used in core, this won't cause performance regression.

repair: use _hexlist() to build revset expression from binary nodes _hexlist() should be efficient than _list().

listkey: display the size of the listkey payload in a debug message This is a useful information to have in general and we already have debug output related to listkeys. I'm planning to play around with massive amount of phases roots and bookmarks so having this data in debug will be very useful. This already got me to spot that one of the Logilab's review repo is exchanging 65KB of phases data during each exchanges.

ssh: test some no-op pull through ssh with --debug It appears that we are never running any wireprotocol operation with a --debug flag. So we are adding some basic testing into 'test-ssh.t'

subrepo: detect issue3781 case earlier so it apply to bundle2 We are doing some strange special casing of phase push when: - the source is a subrepo - the destination is publishing - some changeset are still draft on the destination In that case we do not push phases information (to publish the draft changesets) because it could break simple cycle of 'clone/pull/push' of subrepos. We have to detect this case earlier to have bundle2 respecting it. We change the test to check the behavior for both bundle1 and bundle2.

bundle2: hide bundle2 stream debug under a config flag The old output is very verbose and unsuitable for general debug level. It is however very useful for debugging bundle2 generation or consumption issues. All this verbose ouput is hidden under a 'devel.bundle2.debug' flag.

bundle2: add generic debug output regarding processed interruption If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

bundle2: add generic debug output regarding processed part payload If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

bundle2: add generic debug output regarding processed part If we are about to hide the detailed debug output, we need some generic debug message to replace it.

bundle2: add generic debug output at the end of bundle processing If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

bundle2: add generic debug output regarding processed bundle If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

histedit: fix keep during --continue The --keep option was being serialized to the state file, but it wasn't actually being used when running a histedit --continue. This fixes that.

dispatch: disable demandimport for the --debugger option Something in Python 2.7.9 or so broke the --debugger option with ui.debugger = ipdb. I get the traceback below. There is some apparent confusion with demandimport. This should be disabled anyway for the --debugger option. The debugger must be imported right away, before any other dispatch. There's no benefit in delaying the debugger import. This patch uses the demandimport.deactivated() context manager. Traceback (most recent call last): File "/usr/lib/python2.7/dist-packages/mercurial/dispatch.py", line 121, in _runcatch debugmod = __import__(debugger) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 115, in _demandimport return _hgextimport(_import, name, globals, locals, fromlist, level) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/ipdb/__init__.py", line 16, in <module> from ipdb.__main__ import set_trace, post_mortem, pm, run, runcall, runeval, launch_ipdb_on_exception File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 134, in _demandimport mod = _hgextimport(_origimport, name, globals, locals) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/ipdb/__main__.py", line 29, in <module> if IPython.__version__ > '0.10.2': File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 106, in __getattribute__ self._load() File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 78, in _load mod = _hgextimport(_import, head, globals, locals, None, level) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/IPython/__init__.py", line 45, in <module> from .config.loader import Config File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 132, in _demandimport return _origimport(name, globals, locals, fromlist, level) File "/usr/lib/python2.7/dist-packages/IPython/config/__init__.py", line 16, in <module> from .application import * File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 115, in _demandimport return _hgextimport(_import, name, globals, locals, fromlist, level) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/IPython/config/application.py", line 30, in <module> from IPython.external.decorator import decorator File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 134, in _demandimport mod = _hgextimport(_origimport, name, globals, locals) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/IPython/external/decorator/__init__.py", line 2, in <module> from decorator import * File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 115, in _demandimport return _hgextimport(_import, name, globals, locals, fromlist, level) File "/usr/lib/python2.7/dist-packages/mercurial/demandimport.py", line 47, in _hgextimport return importfunc(name, globals, *args) File "/usr/lib/python2.7/dist-packages/decorator.py", line 240, in <module> 'ContextManager', (_GeneratorContextManager,), dict(__call__=__call__))

hooks: replace if-try-finally with a "with" statement This seems like a textbook case for the new demandimport.deactivated context manager: check if something must be done, do it, and cleanup at the end regardless of exceptions. The diff isn't as bad as it seems. It's just all the whitespace changes due to needing an extra level of indentation. It looks cleaner with `hg diff -w`.

demandimport: define a `deactivated` context manager This can be useful for use in "with" blocks for temporarily disabling demandimport.

largefiles: drop the unused lfcommands._addchangeset()

largefiles: use the convert extension for 'lfconvert --to-normal' The logic in the convert extension is more advanced, supporting extra features like converting revision IDs in 'extras' (e.g. 'amend_source'), supports updating hashes in commit messages, and outputs an SHA map file. Rather than try to duplicate all of that, just use the existing code. Even though the convert extension supports user supplied options like filemap, etc, those features aren't available on the lfconvert interface. Therefore, it is safe to use the filemap mechanism (in memory) to handle the standin -> file rename. The convert extension handles the destination locking for this path. There was a comment in test-lfconvert.t about the hash on rev 5 being different because it was doing a better job than "hg remove" + "hg merge" + "hg commit". It isn't clear to me what was happening or why, but now the hashes match the original repo exactly after a roundtrip, which seems like a good idea. If there really was something beneficial about the previous behavior, perhaps merge can be changed so that everyone benefits. Converting to a largefiles repo still uses the original (limited) lfconvert logic.

bundle2: add generic debug output regarding generated interruption If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

bundle2: add generic debug output regarding generated parts If we are about to hide the detailed debug output, we need some generic debug message to replace it in a concise way.

bundle2: add generic debug output regarding generated bundle If we are about to hide the detailed debug output, we need some generic debug message to replace it in a more compact way.

bundle2: add debug output for part generation The part generation process was lacking a ui object and could not produce debug output. It seems valuable to have some debug output on this part too, especially now that we are planning to be able to hide it in the default --debug output.

bundle2: handle new line in 'indebug' function Now that we have a prefix, it make sense to assume all output will be on a single line.

bundle2: prefix all unbundling debug message with 'bundle2-input:' This make the origin of the message more explicit.

bundle2: introduce a specific function for debug messages while unbundling The bundling process is very verbose, we would like to be able to hide such output behind a configuration flag and have it more explicitly referencing bundle2. The first step is to gather all these messages in a dedicated function.

bundle2: add an informative comment to the capability dict It is fairly easy to get confused by capabilities "missing" from this dict. We make it clear the dict is not the whole story.

merge with stable

bundle2: handle new line in 'outdebug' function Now that we have a prefix, it make sense to assume all output will be on a single line.

bundle2: prefix all bundling debug messages with 'bundle2-ouput:' This makes the origin of the message more explicit.

bundle2: introduce a specific function for bundling debug message The bundling process is very verbose, we would like to be able to hide such output behind a configuration flag and have it more explicitly referencing bundle2. The first step is to gather all these messages in a dedicated function. The same gathering will be later do for debug message issue by unbundling.

parsers: simplify the code computing the phases We recently simplified the code computing the heads of a repo. This patch uses the same simplification for phase computation. We use index_get_parents instead of replicating its code.

parsers: move index_get_parents's declaration higher index_get_parents needs to be used in the phase computation code so we need to move its declaration higher to be able to call it. It cannot be moved any higher than that so we won't have any more patch doing the same thing.

record: add an operation arguments to customize recording ui This patch is part of a series of patches to change the recording ui to reflect the operation currently running (commit, shelve, revert ...). This patch adds a new argument to the recording function to reflect in the UI what operation we are running.

revset: reduce nesting of chained 'or' operations (issue4624) This reduces the stack depth of chained 'or' operations: - from O(n) to O(1) at the parsing, alias expansion and optimization phases - from O(n) to O(log(n)) at the evaluation phase simplifyinfixops() must be applied immediately after the parsing phase. Otherwise, alias expansion would crash by "maximum recursion depth exceeded" error. Test cases use 'x:y|y:z' instead of 'x|y' because I'm planning to optimize 'x|y' in a different way. Benchmarks: 0) 605b1d32c1c0 1) this patch revset #0: 0 + 1 + 2 + ... + 200 0) wall 0.026347 comb 0.030000 user 0.030000 sys 0.000000 (best of 101) 1) wall 0.023858 comb 0.030000 user 0.030000 sys 0.000000 (best of 112) revset #1: 0 + 1 + 2 + ... + 1000 0) maximum recursion depth exceeded 1) wall 0.483341 comb 0.480000 user 0.480000 sys 0.000000 (best of 20) revset #2: sort(0 + 1 + 2 + ... + 200) 0) wall 0.013404 comb 0.010000 user 0.010000 sys 0.000000 (best of 196) 1) wall 0.006814 comb 0.010000 user 0.010000 sys 0.000000 (best of 375) revset #3: sort(0 + 1 + 2 + ... + 1000) 0) maximum recursion depth exceeded 1) wall 0.035240 comb 0.040000 user 0.040000 sys 0.000000 (best of 100)

revset: add helper to build balanced addsets from chained 'or' operations This function will be used by revset.orset() and scmutil.revrange() to reduce the stack depth from O(n) to O(log(n)). We've bikeshed the interface of this function, but we couldn't come to an agreement. So we decided to attempt to make it move forward. marmoute: - new factory function isn't necessary for balanced addsets - addset.__init__ can just recurse, should handle "len(subsets) == 2+" yuja: - want to write all "len(subsets) == 0, 1, 2, 3+" cases in the same function - no recursion in __init__ for cosmetic reason: can't return, can't call __init__ directly I've changed it to a private function so that nobody would be tempted to utilize it.

revset: comment that we can't swap 'or' operands by weight Though the original code did nothing, it tried to optimize the calculation order by weight. But we can't simply swap 'ta' and 'tb' because it would change the order of revisions. For future reference, this patch keeps the modified version of the original code as comment.

parser: add helper to reduce nesting of chained infix operations This will be used to avoid stack overflow caused by chained 'or' operations in revset.

convert: properly pass null ids through .hgtags (issue4678) Mercurial uses tags of null to mark deletions, but convert was silently discarding these because it had no mapping for them. Thus, it was resurrecting deleted tags.

commit: improve --close-branch documentation The confusion about whether this flag closes a whole branch or just a head comes up regularly on IRC, so add some clarification.

patch: add 'extra' argument to makememctx The uncommit command in evolve needs to create memory context with given extra parameters. This patch allows us to do that instead of always giving them an empty value and having to override it afterwards.