cmdutil: add a "changeset.obsolete" label in changeset_printer Until now there were no label to highlight obsolete changesets in log output, only evolution troubles (unstable, bumped, divergent) are supported. We add a "changeset.obsolete" label on changeset entries produced by changeset_printer so that obsolete changesets can be highlighted in log output. This is useful because, unless using a graph log where obsolete changesets have a 'x' marker, there's no way to identify obsolete changesets. And even in graph mode, when working directory's parent is obsolete, we get a '@' marker and we do not see it as obsolete.

fileset: perform membership test against set for status queries Previously, fileset functions operating on status items performed membership tests against a list of items. When there are thousands of items having a specific status, that test can be extremely slow. Changing the membership test to a set makes this operation substantially faster. On the mozilla-central repo: $ hg files -r d14cac631ecc 'set:added()' before: 28.120s after: 0.860s $ hg status --change d14cac631ecc --added 0.690s

worker: flush ui buffers before running the worker a91c6275 introduces flushing ui buffers after a worker finished. If the ui was not flushed before the worker was started, fork will copy the existing buffers to the worker. This causes messages issued before the worker started to be written to the terminal for each worker. We are now flushing the ui before we start a worker and add an appropriate test which will fail before this patch.

chgserver: do not copy configs set by environment variables Config set by environment variables have a source like "$ENVNAME". They should not be copied because they will be recalculated by rcutil.rccomponents.

rcutil: extract duplicated logic to a lambda This simplifies the code a bit.

rcutil: unindent a block Since global _rccomponents is gone, the code could be simplified.

rcutil: do not cache rccomponents The function is only called once except for "hg debugconfig", where it is called twice. So there is no need to cache it. Caching it will cause issues with chgserver. Instead of dropping the cache in chgserver, it seems cleaner to just avoid the cache.

ui: rerun color.setup() once the pager has spawned to honor 'color.pagermode' Otherwise, ui.pageractive is False when color is setup in dispatch.py (without --pager=on), and this config option is ignored.

ui: defer setting pager related properties until the pager has spawned When --pager=on is given, dispatch.py spawns a pager before setting up color. If the pager failed to launch, ui.pageractive was left set to True, so color configured itself based on 'color.pagermode'. A typical MSYS setting would be 'color.mode=auto, color.pagermode=ansi'. In the failure case, this would print a warning, disable the pager, and then print the raw ANSI codes to the terminal. Care needs to be taken, because it appears that leaving ui.pageractive=True was the only thing that prevented an attempt at running the pager again from inside the command. This results in a double warning message, so pager is simply disabled on failure. The ui config settings didn't need to be moved to fix this, but it seemed like the right thing to do for consistency.

color: stop mutating the default effects map A future change will make color.setup() callable a second time when the pager is spawned, in order to honor the 'color.pagermode' setting. The problem was that when 'color.mode=auto' was resolved to 'win32' in the first pass, the default ANSI effects were overwritten, making it impossible to honor 'pagermode=ansi'. Also, the two separate maps didn't have the same keys. The symmetric difference is 'dim' and 'italic' (from ANSI), and 'bold_background' (from win32). Thus, the update left entries that didn't belong for the current mode. This bled through `hg debugcolor`, where the unsupported ANSI keys were listed in 'win32' mode. As an added bonus, this now correctly enables color with MSYS `less` for a command like this, where pager is forced on: $ hg log --config color.pagermode=ansi --pager=yes --color=auto Previously, the output was corrupted. The raw output, as seen through the ANSI blind `more.com` was: <-[-1;6mchangeset: 34840:3580d1197af9<-[-1m ... which MSYS `less -FRX` rendered as: 1;6mchangeset: 34840:3580d1197af91m ... (The two '<-' instances were actually an arrow character that TortoiseHg warned couldn't be encoded, and notepad++ translated to a single '?'.) Returning an empty map for 'ui._colormode == None' seems better that defaulting to '_effects' (since some keys are mode dependent), and is better than None, which blows up `hg debugcolor --color=never`.

pager: do not read from environment variable $PAGER is converted to the pager.pager config item. So it's no longer necessary to read $PAGER in ui.pager().

ui: simplify geteditor Now $EDITOR and $VISUAL will affect ui.editor directly. So it's no longer necessary to test them in ui.geteditor.

debugconfig: list environment variables in debug output Since we print "read config from" for config files, printing environment variables will make it more consistent.

rcutil: let environ override system configs (BC) This is BC because system configs won't be able to override $EDITOR, $PAGER. The new behavior is arguably more rational.

rcutil: add a method to convert environment variables to config items Handling config and environ priorities has been messy. Partially because we don't have config layers - you either get all configs (sys + user), or none. Ideally, environ like $EDITOR, $PAGER should be able to override the system configs "ui.editor", "pager.pager". This patch provides the ability to convert them into config items, so they can be inserted into the middle config layer between system rc and user rc.

rcutil: let rccomponents return different types of configs (API) The next patches will convert environ to raw config items, and insert the config items between systemrcpath and userrcpath. This patch teaches rccomponents to return the type information so the caller could distinguish between "path" and raw config "items".

rcutil: rename rcpath to rccomponents (API)

rcutil: extract rc directory listing logic The logic of listing a ".rc" directory is duplicated in two functions, extract it to a single function to make the code cleaner.

rcutil: split osrcpath to return default.d paths (API) After this change, there are 3 rcpath functions: - defaultrcpath - systemrcpath - userrcpath This will allow us to insert another config layer in the middle.

rcutil: move scmutil.*rcpath to rcutil (API) As discussed at [1], the logic around "actual config"s seem to be non-trivial enough that it's worth a new module. This patch creates the module and move "scmutil.*rcpath" functions there as the first step. More methods will be moved to the module in the future. The module is different from config.py because the latter only cares about data structure and parsing, and does not care about special case, or system config paths, or environment variables. [1]: https://www.mercurial-scm.org/pipermail/mercurial-devel/2017-March/095503.html

statfs: make getfstype() raise OSError It's better for getfstype() function to not suppress an error. Callers can handle it as necessary. Now "hg debugfsinfo" will report OSError.

statfs: rename pygetfstype to getfstype There's no ambiguity now.

statfs: refactor inner function as a mapping from statfs to string The major difference between BSD and Linux is how to get a fstype string. Let's split the longest part of getfstype() as a pure function.

statfs: simplify handling of return value Py_BuildValue() can translate NULL pointer to None.

hghave: use util.getfstype

runtests: change local IP glob pattern from "127.0.0.1" to "$LOCALIP" This is similar to what 348b2b9da703 does. Since 636cf3f7620d has changed "127.0.0.1" to "$LOCALIP". The glob pattern needs update accordingly. It is expected to fix tests running in some BSD jails.

tags: deprecated 'repo.tag' All user are gone. We can now celebrate the removal of some extra line from the 'localrepo' class.

drawdag: use 'tagsmod.tag' instead of 'repo.tag' The former is about to be deprecated.

tags: use the 'tag' function from the 'tags' module in the 'tag' command There is No need to go through the 'repo' object method anymore.

tags: move 'repo.tag' in the 'tags' module Similar logic, pretty much nobody use this method (that creates a tag) so we move it into the 'tags' module were it belong.

tags: move '_tags' from 'repo' to 'tags' module As far as I understand, that function do not needs to be on the local repository class, so we extract it in the 'tags' module were it will be nice and comfortable. We keep the '_' in the name since its only user will follow in the next changeset.

hgweb: filter diff hunks when 'linerange' and 'patch' are specified in filelog

hgweb: add a 'linerange' parameter to webutil.diffs() This is used to filter out hunks based on their range (with respect to 'node2' for patch.diffhunks() call, i.e. 'ctx' for webutil.diffs()). This is the simplest way to filter diff hunks, here done on server side. Later on, it might be interesting to perform this filtering on client side and expose a "toggle" action to alternate between full and filtered diff.

hgweb: handle a "linerange" request parameter in filelog command We now handle a "linerange" URL query parameter to filter filelog using a logic similar to followlines() revset. The URL syntax is: log/<rev>/<file>?linerange=<fromline>:<toline> As a result, filelog entries only consists of revision changing specified line range. The linerange information is propagated to "more"/"less" navigation links but not to numeric navigation links as this would apparently require a dedicated "revnav" class. Only update the "paper" template in this patch.

shelve: add logic to preserve active bookmarks This adds an explicit active-bookmark-handling logic to shelve. Traditional shelve handles it by transaction aborts, but it is a bit ugly and having an explicit functionality seems better.

metadataonlyctx: speed up sanity check Previously the sanity check will construct manifestctx for both p1 and p2. But it only needs the "manifest node" information, which could be read from changelog directly.

revset: factor out linerange processing into a utility function Similar processing will be done in hgweb.webutil in forthcoming changeset.

hgweb: add a "patch" query parameter to filelog command Add support for a "patch" query parameter in filelog web command similar to --patch option of `hg log` to display the diff of each changeset in the table of revisions. The diff text is displayed in a dedicated row of the table that follows the existing one for each entry and spans over all columns. Only update "paper" template in this patch.

hgweb: handle "parity" internally in webutil.diffs() There's apparently no reason to have the "parity" of diff blocks that webutil.diffs() generates coming from outside the function. So have it internally managed. We thus now pass a "web" object to webutil.diffs() to get access to both "repo" and "stripecount" attribute.

largefiles: add lfile argument to updatestandin() for efficiency (API) Before this patch, updatestandin() takes "standin" argument, and applies splitstandin() on it to pick out a path to largefile (aka "lfile" or so) from standin. But in fact, all callers already knows "lfile". In addition to it, many callers knows both "standin" and "lfile". Therefore, making updatestandin() take only one of "standin" or "lfile" is inefficient.

largefiles: use strip() instead of slicing to get rid of EOL of standin This slicing prevents from replacing SHA-1 by another (= longer hash value) in the future.

largefiles: rename local variable appropriately repo['.'] is called not as "working context" but as "parent context". In this code path, hash value of current content of file should be compared against hash value recorded in "parent context". Therefore, "wctx" may cause misunderstanding in this case.

largefiles: avoid redundant loop to eliminate None from list Before this patch, this code path contains two loops for m._files: one for replacement with standin, and another for elimination of None, which comes from previous replacement ("standin in wctx or lfdirstate[f] == 'r'" case in tostandin()). These two loops can be unified into simple one "for" loop.

largefiles: avoid meaningless changectx looking up Logically, "repo[ctx.node()]" should be equal to "ctx". In addition to it, this redundant code path is repeated "len(match.m_files)" times.

largefiles: avoid redundant changectx looking up at each repetitions These code paths look up changectx at each repetitions, even though the changectx key isn't changed while loop.

largefiles: omit updating newly added standin at linear merging Updating standin for newly added largefile is needed, only if same name largefile exists in destination context at linear merging. In such case, updated standin is used to detect divergence of largefile at overridefilemerge(). Otherwise, standin doesn't have any responsibility for its content (usually, it is empty).

largefiles: reuse hexsha1() to centralize hash calculation logic into it This patch also renames argument of hexsha1(), not only for readability ("data" isn't good name for file-like object), but also for reviewability (including hexsha1() code helps reviewers to confirm how these functions are similar). BTW, copyandhash() has also similar logic, but it can't reuse hexsha1(), because it writes read-in data into specified fileobj simultaneously.

py3: prove second commit works Finally it works.

py3: fix manifestdict.fastdelta() to be compatible with memoryview This doesn't look nice, but a straightforward way to support Python 3. bytes(m[start:end]) is needed because a memoryview doesn't support ordering operations. On Python 2, m[start:end] returns a bytes object even if m is a buffer, so calling bytes() should involve no additional copy. I'm tired of trying cleaner alternatives, including: a. extend memoryview to be compatible with buffer type => memoryview is not an acceptable base type b. wrap memoryview by buffer-like class => zlib complains it isn't bytes-like

crecord: use ProgrammingError

transaction: use ProgrammingError

bundle2: use ProgrammingError

merge: use ProgrammingError

repair: use ProgrammingError

py3: abuse r'' to preserve str-ness of literals passed to __setattr__()

py3: fix slicing of byte string in revlog.compress() I tried .startswith('\0'), but data wasn't always a bytes nor a bytearray.

py3: use bytes() to cast to immutable bytes in changelog.appender.write()

py3: use bytes() to cast to immutable bytes in pure.bdiff.bdiff()

bdiff: drop support for array.array argument from pure.bdiff.bdiff() Thanks to 2a18e9e6ca43, we no longer pass array.array('c') object to bdiff().

histedit: test that an aborted histedit can be rerun (with obsolete) In the future, this should help catching issue as the one introduced in 6f0b7475cf9a.