As of Python 3, decimal notations of octal values for permission modes
are no longer permitted and will result in a `SyntaxError` exception
(`invalid token`).
Using the proper octal notation which is also Python 2.7 compatible will
fix this issue.
--HG--
extra : rebase_source : 2e897c51f04ad0ee69071f84b98df224f3af72d3
Mingw python has a different os.path setup from native python, and has
os.sep and os.altsep reversed. In that case, the normsep function was
doing the wrong thing, leading to all sorts of problems.
While fixing this, also ensure the corresponding unit test covers this
peculiarity, even when running under the native win32 python.
--HG--
extra : rebase_source : 8fb18e0d4dc669c1d7e069f73fc44c22d419d43c
I have a need to create tar archives deterministically
and reproducibly. Since we already have similar functionality in
mozpack for producting zip/jar archives, I figured it made sense for
this functionality to live in mozpack.
I made the functionality as simple as possible: we only accept
files from the filesystem and the set of files must be known in
advance. No class to hold/buffer state: just a simple function
that takes a mapping of files and writes to a stream.
MozReview-Commit-ID: If0NTcA7wpc
--HG--
extra : rebase_source : 9cbea36347ba2840dd5bff9dffefd994a73a0725
This will be needed to teach artifact builds to extract files from
omni.ja files whose content is loaded into memory (from a tar
archive).
MozReview-Commit-ID: LH2HkKx5Zj3
--HG--
extra : rebase_source : 7f6b176f0ef9fc87889151d1d02da62de8a455d8
extra : source : 266928b5a7615fa054c70adf0f649cbb3f085e8d
Generally speaking, the configuration needs forward-slashes in paths.
We might as well make it hard(er) to set configuration items with
backslash separators on Windows by exposing mozpath.* functions in place
of os.path functions. The downside is that functions explicitly
importing os will still get the real os.path functions.
As previous measurements have shown, creating/appending files
on Windows/NTFS is slow because the CloseHandle() Win32 API takes
1-3ms to complete. This is apparently due to a fundamental issue
with NTFS extents. A way to work around this slowness is to use
multiple threads for I/O so file closing doesn't block execution
as much.
This commit updates the file copier to use a thread pool of 4
threads when processing file copies. Additional threads appear
to have diminishing returns.
On my i7-6700K, this reduces the time for processing the tests install
manifest (24,572 files) on Windows from ~22.0s to ~12.5s in the best
case.
Using the thread pool globally resulted in a performance regression
on Linux. Given the performance sensitivity of manifest copying,
I thought it best to implement a slightly redundant non-Windows
branch to preserve performance. For the record, that same machine
running Linux is capable of processing nearly the same install
manifest (24,616 files) in ~2.2s in the best case.
MozReview-Commit-ID: B9LbKaOoO1u
--HG--
extra : rebase_source : e9fee3861a70e1da9d18448f8435f4bd3e28c647
This is two straightforward optimizations in FileCopier: avoiding a redundant iteration
over the directory structure to find destination files (which includes a
call to normpath) and avoiding redundant calls to determine directories to preserve
when remove_unaccounted is not specified (which include a call to dirname).
Running a no-op install of _tests with this patch results in a reduction of about
25,000 calls to normpath and remove about 220,000 calls to dirname, resulting in
an overall speedup of 10-20%.
--HG--
extra : commitid : 8nyTo489q8X
Future improvements to process_install_manifest's --track option will require
adding data in the tracking dump that uses an install manifest form, and I don't
want e.g. switching branches or bisection to require to clobber in order to do the
right thing, so this change future-proofs the install manifest reader.
The default behavior for a FileCopier's copy is to remove all the files and
directories in the destination that aren't in its registry.
The remove_unaccounted argument can be passed as False to disable this
behavior.
This change adds another possibility, where remove_unaccounted may be a
FileRegistry, in which case only the files in that registry are removed.
This allows to e.g. only remove files that were copied from a previous
FileCopier.copy, leaving aside files that were in the destination for some
other reason.
Future improvements to process_install_manifest's --track option will require
adding data in the tracking dump that uses an install manifest form, and I don't
want e.g. switching branches or bisection to require to clobber in order to do the
right thing, so this change future-proofs the install manifest reader.
The default behavior for a FileCopier's copy is to remove all the files and
directories in the destination that aren't in its registry.
The remove_unaccounted argument can be passed as False to disable this
behavior.
This change adds another possibility, where remove_unaccounted may be a
FileRegistry, in which case only the files in that registry are removed.
This allows to e.g. only remove files that were copied from a previous
FileCopier.copy, leaving aside files that were in the destination for some
other reason.
FlatFormatter, JarFormatter and OmniJarFormatter all, in some way, deal
with different pieces of the package being handled differently.
Instead of each of them dealing with their different pieces in some subtly
different way, instead, introduce a new base package formatter class that
will handle it for all of them.
Use this new PiecemealFormatter for the FlatFormatter.
Only directories containing chrome manifests are given as base to formatters,
but there can still be files given outside the bases, like, on mac builds,
all files in Content/MacOS, or Content/Info.plist, whereas chrome manifests
are under Content/Resources.
There is a lot of repetition across its various tests, and we're going to add
some more in a subsequent change, so it is desirable to make it a less
repetitive task.
This function was found to be a little slow while profiling due to repeated calls to
mozpath.dirname. This patch speeds up the function replacing dirname with string manipulation
(these paths are already normalized), by caching results on the basis of directory,
and converting from iteration to recursion to increase use of the cache.
This commit speeds up the "install tests" step run as a part of the build and running
tests by ~10% on a fast linux laptop.
--HG--
extra : commitid : HdYkcXQ2ezQ
An upcoming commit will introduce a caller that doesn't want the maximum
compression level. This commit introduces arguments to control the
compression level inside written archives.
--HG--
extra : commitid : KkDso3hB2QG
extra : rebase_source : 8fd05aeae5c3555e1169eac6656d584007cd0739
Previously, we always skipped over files beginning with a ".". This
commit adds an option to include them.
This is needed to support test package generation via Python / mozpack.
--HG--
extra : commitid : 4pmEpukVX0s
extra : rebase_source : 31599230ce344b9be815b3a457cc8a7c6d8e5301
The hglib Mercurial finder was nice. But it is somewhat slow, as it
involves a separate Mercurial process.
This commit introduces a native Mercurial finder that speaks directly
to a Mercurial repository instance. It is significantly faster.
The new code is isolated to its own file because it imports Mercurial
code, which is GPL.
--HG--
extra : commitid : 8CzDFt3lQmx
extra : rebase_source : 1540ad20d795086926b316612062e2a1f10c4958
The moz.build reader uses absolute paths when referencing moz.build
files. *Finder classes expect paths to be relative to a base. When we
switched the reader to use *Finder instances for I/O, we simply provided
a default FileFinder based at the filesystem root. This was quick and
easy. Things worked.
Unfortunately, that solution isn't ideal because not all *Finder
instances can accept absolute paths like that. The
MercurialRevisionFinder is one of them.
Changing the moz.build reader to talk in terms of relative paths is a
lot of work. While this would be ideal, it is significantly easier to
defer the problem until later and hack up MercurialRevisionFinder to
accept absolute paths. This commit does exactly that.
Bug 1171069 has been filed to track converting BuildReader to relative
paths.
--HG--
extra : commitid : 2JmzOBldBLy
extra : rebase_source : a8af6ce88dd9e2b98f131c92b45c3ece852e13d2
Now that moz.build files use finders for I/O, we can start reading
moz.build data from other sources.
Version control is essentially a filesystem. We implement a finder
that speaks to Mercurial to obtain file data. It is able to obtain
file data from a specific revision in the repository.
We use the hglib package (which uses the Mercurial command server) for
speaking with Mercurial. This adds overhead compared to consuming the
raw Mercurial APIs. However, it also avoids GPL side-effects of
importing Mercurial's Python modules.
Testing shows that performance is good but not great. A follow-up
commit will introduce a GPL licensed Mercurial finder. For now, get
the base functionality in place.
--HG--
extra : commitid : BkwaQOW9MiR
extra : rebase_source : 915d6015317ccc79c228a76eed861d9f43e2fd17
Passing raw file handles around is a bit dangerous because it could lead
to leaking file descriptors. Add a read() method that handles the simple
case of obtaining the full contents of a File instance.
This should ideally be a method on BaseFile. But this would require
extra work and isn't needed. So we've deferred it until bug 1170329.
--HG--
extra : commitid : 82qw76XmpjC
extra : rebase_source : 422b16c5a3b1577f080097925aeaeb560aa3e798
Today, the *Finder classes are optimized for doing matching and
returning multiple results. However, sometimes only looking for a
single file is wanted.
This commit implements the "get" method on all the *Finder classes.
It returns a BaseFile or None.
FileFinder._find_file has been refactored into FileFinder.get
to avoid code duplication.
--HG--
extra : commitid : K9It2ZJ3Rbo
extra : rebase_source : a56f8f70aa1902d26373a7196eae2847cce653d3
A host elfhack binary is only built when there is a compile environment.
--HG--
extra : rebase_source : 4f1da429c581dfd81cbe3d5164c7586066cf6e79
extra : amend_source : f26fe7f3b44291f38459a81b9ff31bd6dbd220aa
extra : histedit_source : 93c628a500c66c46d522bfe678500bf5b5bf0de9
This avoids duplicating the logic from SimplePackager to find base
directories, and fixes some cases where the l10n repack code wouldn't
find them properly.
Bug 910660 added a consistency check that rejects cases where a manifest
inside a directory detected as being the base of an addon is included from
outside the addon. Unfortunately, this triggered false positives when
a manifest is included from within the addon, but just happens to be at
the top-level of that addon.
This avoids duplicating the logic from SimplePackager to find base
directories, and fixes some cases where the l10n repack code wouldn't
find them properly.
Some file types, such as XPTFile, read their associated data during the
copy. In the case of XPTFiles, several original files are linked together
in one destination file. When a FileCopier is used in-place, those
original files are removed before XPTFile.copy is invoked, so XPTFile.copy
fails.
Generally speaking, there are many other reasons why FileCopier can fail
in-place, but the only active use of this mode is l10n repack code, which
actually doesn't do much of the dangerous uses. However, it can end up
linking XPTFiles for some reason, which fails for the reasons given above.
Back when mozpack.path was added, it was used as:
import mozpack.path
mozpack.path.func()
Nowadays, the common idiom is:
import mozpack.path as mozpath
mozpath.func()
because it's shorter.
$ git grep mozpath\\. | wc -l
423
$ git grep mozpack.path\\. | wc -l
123
This change was done with:
$ git grep -l mozpack.path\\. | xargs sed -i 's/mozpack\.path\./mozpath./g'
$ git grep -l 'import mozpack.path$' | xargs sed -i 's/import mozpack.path$/\0 as mozpath/'
$ (pat='import mozpack.path as mozpath'; git grep -l "$pat" | xargs sed -i "1,/$pat/b;/$pat/d")
The regular expression cache for mozpack.path.match was keyed off the
original pattern. However, that variable was mutated as part of the
function and the mutated result was subsequently stored as the cache
key. This effectively resulted in a 0% cache hit rate.
On some tests being written for bug 1132111 which involve a full
filesystem traversal for moz.build files and subsequent execution of
those files, the following timings are indicative of the impact of this
patch.
Before:
real 16.082s
user 14.760s
sys 1.318s
After:
real 6.345s
user 5.085s
sys 1.257s
--HG--
extra : rebase_source : caf4a9f37fda9b43b444059f647535e1b085d422
mopack.BaseFile.copy() performs a generic read/write file copy. Windows
has an explicit CopyFile() call that tests have shown to be
significantly faster. Let's use that instead via the magic of ctypes.
This is a straight copy from
a878bf0ba0
paired with a tiny change to use the new quote_chars option.
--HG--
extra : rebase_source : 75d604ffafc7062c663bca4242af35546d2c1e3a
The forward slash appears to be the standard path separator in zip/JAR
files. Accept back slashes when adding paths to a JAR.
--HG--
extra : rebase_source : bd94eab36b347006e65952d99b53dd397e2ca758
extra : amend_source : 2cefd887d8bb5d989fafb398a3464429ac376e2e
The install manifest processor starts with an empty InstallManifest and
uses |= to "concatenate" instances. It became pretty obvious when
developing some patches that add more preprocessed files to install
manifests that the source install manifest dependency was getting
lost during the |= operation. This patch fixes it.
The solution is not ideal performance wise. But slightly worse
performance (only after config.status, however) is better than
clobbers.
A test has been added to ensure this doesn't regress.
--HG--
extra : rebase_source : 848aebbbc935ce2bca2d3fcc85d1df534e734e0d
Smart xpt linking will keep around [scriptable] interfaces and anything
those interfaces depend on. Modify the tests that deal with xpt linking
so they use [scriptable] interfaces, ensuring that the tests continue to
work in the face of smarter linkers.
This already raised if the order was [foo, foo/bar]. But it didn't
prevent adding [foo/bar, foo].
The only sub-classes of FileRegistry are FileCopier and Jarrer.
FileCopier.copy threw in the previously unhandled case: the order of
creation is the same as the order of addition, so that foo is created
after foo/bar.
A zip file index can contain both foo and foo/bar. I don't think we
should rely on this property in our use of Jarrer, but if we already do,
I guess we need to move these guards into FileCopier. Let's hope that's
not the case!
(For the record: On my Mac OS X system, unzipping such a zip file
prompts the user for what to do, depending on the order of the entries
in the zip index.)
This patch adds pattern matching entries to install manifests. We store
metadata necessary to construct a pattern match at a later point in
time. When we convert the install manifest to a file registry, we
resolve the patterns using FileFinder.
The build config logic has been updated to store support-files values as
pattern entries. This should resolve the clobber needed issue and make
the local development experience more pleasant as well.
--HG--
extra : amend_source : 3fe659f7ad6930ef54316b5babac6b83bee240af
aleth
Yen Chi Hsuan
Mike Hommey
Henrik Skupin
Wes Kocher
Chris Manchester
Elias Probst
Gregory Szorc
Chris AtLee
Carsten "Tomcat" Book
Ms2ger
Nick Alexander
Myk Melez
Gijs Kruitbosch
Ted Mielczarek
Ryan VanderMeulen
Mike Hommey
Philipp Kewisch
Nathan Froyd
Bill McCloskey
Ehsan Akhgari
Brian O'Keefe