Some compilers on some platforms by default #define some of the values
we're using in the source we use in get_compiler_info(). Namely,
mingw-gcc #defines WINNT by default, and the WINNT in the source is then
replaced by 1, breaking the check.
The C preprocessor, fortunately, doesn't expand macros inside C strings.
So instead of `%KERNEL WINNT`, we output `%KERNEL "WINNT"`, and strip
out the double quotes. For good measure, we do this for all values in
the source used in get_compiler_info().
--HG--
extra : rebase_source : dd4cc2b8c3bf0cb508b09598706b74ccc12162be
.mozbuild/last_log.json is opened by the process doing the deleting.
On Windows, you can't delete an opened file.
So stop clobbering the .mozbuild directory.
This directory only holds non-essential artifacts from mach/build
invocations. I don't think keeping it around will break anything.
MozReview-Commit-ID: 9b0AC2ywAAZ
--HG--
extra : rebase_source : 9236e3e282d1db2595e8b36353008bc98e2ae6cf
Some substs values are lists, and some CONFIGURE_SUBST_FILES use them,
and don't expect the substitution to use a pythonic expression of the
value. So serialize those lists.
--HG--
extra : rebase_source : 848962c326236607609bca2b92180c8f6f4ad079
The "remove_objdir" method has been rewritten to support partial
clobber. It still defaults to full clobber. But the "full" argument
can be passed as False to limit to a partial clobber where currently
the "msvc" directory (contains Visual Studio files) is not clobbered.
On Windows, there might be a regression with this change because
we'll be invoking N winrm.exe processes and new processes have
a non-trivial overhead on Windows. However, it is hopefully unlikely
that new processes are more overhead than deleting hundreds of thousands
of files.
MozReview-Commit-ID: 7yeMttztwic
--HG--
extra : rebase_source : 646992be199e1059f0b09cf8bed3334085293fe0
My Python editor told me rmtree has been marked as deprecated. Use
mozfile.remove instead.
MozReview-Commit-ID: FYkXk1fnxvC
--HG--
extra : rebase_source : df392e8832e52501950f09acda208943c9e3d707
This required implementing a utility function to resolve the binary
type. I used GetBinaryTypeW via ctypes because this seems the fastest.
I arbitrarily limited the function to testing 32-bit and 64-bit Windows
executables because hopefully those are the only executables we'll
ever encounter. We can expand the binary detection later, if needed.
This includes support for running on non-Windows platforms.
MozReview-Commit-ID: CYwyDWQrePc
--HG--
extra : rebase_source : 8fd7ca7f253d9e9e18d64784652a5ff934ad2272
And with indentation so it is easier for humans to read.
MozReview-Commit-ID: Kkd6vmfLNUf
--HG--
extra : rebase_source : 9f2b4ad2223f361dd6dd1ec9cdda71bd5a8560e3
Until now, HAVE_64BIT_BUILD was entirely determined by a compiler check.
But we didn't run the check on e.g. artifact builds, while relying on
its result for some non-compilation related things, leading to subtle
discrepancies.
This changes the configure check to derive HAVE_64BIT_BUILD from bitness
determined by the target CPU, and double checked with a compiler check.
--HG--
extra : rebase_source : 5dc0cf2369ed4457bdd9a15736a70265a771d919
Currently, it returns either None or the contents of the compiler's stdout,
which is always expected to be an empty string, and is not very useful. So
instead, return True in the latter case.
--HG--
extra : rebase_source : ee69cdeab38d27178ce759591fb394da65e694ac
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
The base compiler check in python configure does some preprocessing,
which ensures the compiler works to some extent. Autoconf used to have
a more complete test, doing a compile/link. We do have plenty of tests
afterwards that do that anyways, but it's better if we fail early if
the toolchain fails somehow.
This refactors try_compile such that the *_compiler variable themselves
can be used to trigger compiler tests. Eventually, we'll want something
similar for preprocessing and possibly other invocations.
This also removes similar tests from build/autoconf/toolchain.m4.
--HG--
extra : rebase_source : c60d1d6e39b6bd2a377516687affd9b8932ebc12
While on automation, there is no MSVC to find through the registry, the
story is different on local builds, and that can interfere with tests.
Specifically, it breaks test_toolchain_configure.py because it's not
expecting the registry to provide a valid path to an almost valid
compiler, and then fails because that compiler doesn't match the
expected target CPU.
And because build/moz.configure/toolchain.configure also affects the PATH
environment variable, subsequent tests end up failing even earlier
because executing the empty mozconfig with the modified environment then
fails because of the unicode value of PATH.
This change implements enough of _winreg to make the get_registry_values
function return nothing.
--HG--
extra : rebase_source : f70e40ddabaed1197f6cddce67832bb112f1969d
Back when it was added, it was used, but it is not anymore, outside
test_base.py.
--HG--
extra : rebase_source : f0b9a4dab2985e89e9950eda774ae853c7de764c
We've been reading the mozconfig in MozbuildObject.from_environment to
check whether the mozconfig topobjdir matches the detected topobjdir.
Since bug 1278415, everything using the buildconfig python module now
calls MozbuildObject.from_environment, which reads the mozconfig. A lot
of things to that during the build. But none of them actually need the
data from the mozconfig, and the topobjdir match test has been breaking
things randomly on multiple occasions.
The topobjdir match test, however, really only needs to happen once:
when a mach command starts. So we can move the test to MachCommandBase,
where it belongs, and anything actively using MozbuildObject.mozconfig
will have the mozconfig read, but everything else won't.
On a Linux64 opt build, this brings down the number of times the
mozconfig is read during `mach build` from 979 to 9.
--HG--
extra : rebase_source : 6b340f1fcf73a3c3987033c37f8f14ef06a44f04
This patch is really two separate changes.
The first change is that rust crates are large, standalone entities that
may contain multitudes of source files. It therefore doesn't make sense
to keep them in SOURCES, as we have been doing. Moving to use cargo
will require a higher-level approach, which suggests that we need a
different, higher-level representation for Rust sources in the build
system.
The representation here is to have the build system refer to things
defined in Cargo.toml files as the entities dealt with in the build
system, and let Cargo deal with the details of actually building things.
This approach means that adding a new crate to an existing library just
requires editing Rust and Cargo.toml files, rather than dealing with
moz.build, which seems more natural to Rust programmers. By having the
source files for libraries (and binaries in subsequent iterations of
this support) checked in to the tree, we can also take advantage of
Cargo.lock files.
The second is that we switch the core build system over to building via
cargo, rather than invoking rustc directly.
We also clean up a number of leftover things from the Old Way of doing
things. A number of tests are added to confirm that we'll only permit
crates to be built that have dependencies in-tree.
rlibs are going to be less important once we start building with cargo:
the focus will move to crates instead, so that's what we should call the
moz.build object.
Through an oversight, we listed librul.a twice when linking libxul: once
as part of the "objects" we were linking, and once as a static library.
This duplication is unnecessary and would cause problems later when we
try to generate librul.a via cargo, as cargo will put it someplace
different from where we expect and the two names will conflict. Let's
have rules.mk be the single source of truth for how librul.a is named,
and then the code to link libxul can simply refer to that name.
This patch is really two separate changes.
The first change is that rust crates are large, standalone entities that
may contain multitudes of source files. It therefore doesn't make sense
to keep them in SOURCES, as we have been doing. Moving to use cargo
will require a higher-level approach, which suggests that we need a
different, higher-level representation for Rust sources in the build
system.
The representation here is to have the build system refer to things
defined in Cargo.toml files as the entities dealt with in the build
system, and let Cargo deal with the details of actually building things.
This approach means that adding a new crate to an existing library just
requires editing Rust and Cargo.toml files, rather than dealing with
moz.build, which seems more natural to Rust programmers. By having the
source files for libraries (and binaries in subsequent iterations of
this support) checked in to the tree, we can also take advantage of
Cargo.lock files.
The second is that we switch the core build system over to building via
cargo, rather than invoking rustc directly.
We also clean up a number of leftover things from the Old Way of doing
things. A number of tests are added to confirm that we'll only permit
crates to be built that have dependencies in-tree.
rlibs are going to be less important once we start building with cargo:
the focus will move to crates instead, so that's what we should call the
moz.build object.
Through an oversight, we listed librul.a twice when linking libxul: once
as part of the "objects" we were linking, and once as a static library.
This duplication is unnecessary and would cause problems later when we
try to generate librul.a via cargo, as cargo will put it someplace
different from where we expect and the two names will conflict. Let's
have rules.mk be the single source of truth for how librul.a is named,
and then the code to link libxul can simply refer to that name.
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
This never was a problem since auto-logging was broken on Windows, but
having mach clobber auto-log is a problem on Windows since it tries to
remove the log file while itself has it open, which fails.
--HG--
extra : rebase_source : 6da9f3e6148c201eade01d4598ce9c4becd051a0
We were relying on the log manager's terminal to know whether we're
running on a terminal to turn on auto-logging, but the log manager's
terminal is always None on Windows because blessings imports curses,
which doesn't work on Windows.
So instead, just use a plain os.isatty() call.
The same problem also applies to the show-log command to trigger the
pager (less). At the same time, fix the environment setting for LESS,
as on Windows, unicode literals are not allowed in the environment.
--HG--
extra : rebase_source : 1462849608c8cb86afcb080184efb50caf6d2f9e
The best kind of patch: bulk deletion. This removes two separate but
similar build flags, and an unsupported integration piece. The first
packaged Fennec's resources into an ill-defined GeckoView archive; the
second built on the first to produce an Android ARchive for external
consumption. Neither of these artifacts are supported or have
consumers; in fact, they mislead potential consumers, since they're
known to be broken. The Gradle build work under the --with-gradle
flag, and significant follow-up, is the path forward if Mozilla wants
to invest in packaging GeckoView on Android for external consumption.
That is, rather than hacking together AAR files, we'll use the
well-supported Gradle mechanisms for building and publishing such
libraries.
MozReview-Commit-ID: 4Z1jJ8z0cyJ
--HG--
extra : rebase_source : b8e65f39c286313fe8963bf3832d9b6977ef188f
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
Prevailing style in the file for when read_topsrcdir is being tested for
exception-throwingness is to omit the assignment, so let's make
everything consistent.
And for GCC and clang, try to see if adding -m32, -m64 or --target
works if they don't.
--HG--
extra : rebase_source : 874bc2404a5ccc48e938bc7d9b2fe67ba625cb3e
This removes the unnecessary setting of c-basic-offset from all
python-mode files.
This was automatically generated using
perl -pi -e 's/; *c-basic-offset: *[0-9]+//'
... on the affected files.
The bulk of these files are moz.build files but there a few others as
well.
MozReview-Commit-ID: 2pPf3DEiZqx
--HG--
extra : rebase_source : 0a7dcac80b924174a2c429b093791148ea6ac204
This makes building on msys2 easier since its pip is broken.
MozReview-Commit-ID: 1hQHeu3BKOd
--HG--
extra : rebase_source : 5447d96893a502225980d1dab7b4f89b888ad661
This adds the 'xpcshell-test' command to the mach environment found in the test
package. This will allow developers to easily run xpcshell after checking out
and interactive worker.
MozReview-Commit-ID: fBAbfuG5XQ
--HG--
extra : rebase_source : 71077c9142f33843ed87d4bc4617a780f775939b
Before bin_path would detect msys2 as if it were a windows system which would cause a virtualenv problem since the binaries should go in a /bin folder not a /Scripts folder for msys2.
Based on a patch by :vlad 89b21efe93
MozReview-Commit-ID: IyAukszHGmN
--HG--
extra : rebase_source : 9ada9cf61b76c85cf483ad8212c95d17070a7025
Technically, this includes the mingw64 version of make, but it is still called mingw32-make when installed from pacman with msys2.
This is a necessary step to make Firefox build on a pure msys2 environment.
Also, unlike mozmake, because this make is installed from a package manager. It can be automatically updated.
MozReview-Commit-ID: EAN6xmIgYWd
--HG--
extra : rebase_source : 43fe0d4b35e91f1143ce2ea1d20f5f3c5f1f7c86
The mozconfig detection logic has bitten us on many occasions in the
past. The following changes are made to tentatively improve the
situation:
- The API is modified such that autodetection of the mozconfig has
to be a conscious decision made by the caller, and not triggered
any time there is no mozconfig given, which could be a conscious
decision of the opposite.
- mozinfo.json now stores the actual mozconfig (or lack thereof) used
during configure.
--HG--
extra : rebase_source : c7a632afd414f25daf7bbe7e1a66c3736c26e039
Per froydnj in bug 1186064 comment #23, "it makes sense to proceed with removing
MSVC 2013 support." This commit does that.
We also go a step further and require VS2015 Update 2 instead of just
update 1. This temporarily brings us down to just 1 officially supported
Visual Studio version. However, VS2015u3 was just released and is
unofficially supported.
Since MOZ_CRT is no longer set, references to it have been removed.
MozReview-Commit-ID: 8MUR6qLzQA5
--HG--
extra : rebase_source : 8f5061080a3d56dd484f9be03649fb65f0145f67
Per froydnj in bug 1186064 comment #23, "it makes sense to proceed with removing
MSVC 2013 support." This commit does that.
We also go a step further and require VS2015 Update 2 instead of just
update 1. This temporarily brings us down to just 1 officially supported
Visual Studio version. However, VS2015u3 was just released and is
unofficially supported.
Since MOZ_CRT is no longer set, references to it have been removed.
MozReview-Commit-ID: 8MUR6qLzQA5
--HG--
extra : rebase_source : 22ab4f47661ead4995d0c5261104abfb02b82aa2
Since it conflicts with moz.configure which resolves shortnames since
parts of the build system don't like paths with spaces.
MozReview-Commit-ID: nJkRV4Nafo
--HG--
extra : rebase_source : 9f2db54ec38eb0f05b478b445c0a883ca69b0127
This also fixes the issue of processing the artifacts twice in some
situations (bug 1275673). Note that the artifact download no longer
happens when a specific target is passed to 'mach build'.
MozReview-Commit-ID: Ktys6u3r1kG
The buildconfig module predates MozbuildObject.from_environment, and
it's about time to start factoring things out such that we only have
one way to get config.status data. This is step 1: making the
buildconfig module use MozbuildObject.from_environment.
Eventually, we'll want to remove the buildconfig module uses everywhere.
The topobjdir-finding logic in mozbuild relies on MOZ_CURRENT_PROJECT
being set, and it's currently only set when going through client.mk.
On automation, during universal builds, make check is invoked directly
in one of the objdirs, so MOZ_CURRENT_PROJECT is not set. We've had
other similar problems in the past. Ensuring MOZ_CURRENT_PROJECT is set
in the objdir itself should reduce the risk of other such issues in the
future.
Historically, a mozinfo for js standalone build has not been necessary,
but with the move towards a) having things work with mach and b)
buildconfig using the MozbuildObject.from_environment in next patch,
mozinfo becomes necessary (it's required for
MozbuildObject.from_environment to find the directory is an objdir).
Interestingly, hazard builds do both a js standalone build and a desktop
Firefox build at the same time, both of which are done with MOZCONFIG
set in the environment... with the Firefox mozconfig. The result of now
writing a mozinfo for js standalone builds is that in that case, they
end up with a reference to the mozconfig, which the build system then
reads, and finds a MOZ_OBJDIR, which then doesn't match the js
standalone build objdir. So for those builds, reset MOZCONFIG.
Killing the sccache background daemon is part of postflight_all, but in
the current setup, postflight_all happens at the end of a "normal" build,
but we run automation build steps after that.
What happens then is that more compilations happen (gtests), which start
sccache again, but there's nothing to kill sccache again once this is
all done.
Now that the OSX universal builds postflight is gone, it is not
necessary for postflight_all to happen before the automation build steps.
So ensure postflight_all scripts happen last.
The downside of this change is that this now prevents sccache.log from
being uploaded, but we should probably send processed data to the graph
server instead.
Killing the sccache background daemon is part of postflight_all, but in
the current setup, postflight_all happens at the end of a "normal" build,
but we run automation build steps after that.
What happens then is that more compilations happen (gtests), which start
sccache again, but there's nothing to kill sccache again once this is
all done.
Now that the OSX universal builds postflight is gone, it is not
necessary for postflight_all to happen before the automation build steps.
So ensure postflight_all scripts happen last.
The downside of this change is that this now prevents sccache.log from
being uploaded, but we should probably send processed data to the graph
server instead.
Now that the VisualStudio backend will no-op if nothing has changed, it
should be safe to always run this backend.
On first run, backend generation takes ~3.5s on my machine. On subsequent run,
it takes ~1.5s. Wall time for a no-op config.status is now ~15.7s. We could like
make the Visual Studio backend faster by not writing so many project files.
But this would require consolidating libraries in moz.build files. And that's
out of scope for this change.
We drop the check for MSVS_VERSION because it won't always be defined on
MinGW/GCC builds. We simply default to "2015" if it isn't set.
MozReview-Commit-ID: 5W38HMGmcuV
--HG--
extra : rebase_source : 302d76277058819c115f3c2518b8cb2067971950
extra : source : 408319d87eacb28848efeab0346eb815440adade
There is probably a way to dynamically retrieve the version. But rather
than take the chance we'd query the wrong thing, let's just parse the
version that Visual Studio writes to the solution file when saving it and
use it.
With this change, generating the VisualStudio build backend should not change
any files unless the build config has changed. This means we can generate
Visual Studio files at will without causing Visual Studio to complain
about the solution and other files changing and needing reloading.
MozReview-Commit-ID: 1udZ72SLEzP
--HG--
extra : rebase_source : d15bff5b30a021dc1180e48fb7bb0d6c84871b69
Visual Studio will write this variable to an ancient Visual Studio
version (2010 I believe) if we don't specify it. Explicitly write the
variable to the minimum Visual Studio version we support.
MozReview-Commit-ID: 8Y0im48OM2G
--HG--
extra : rebase_source : 7b1f4cfe778c6258dce068b2aec8b6acd8e85102
Upon further examination, VS2013 and VS2015 share the same file format
version. They also write the internal version number in a comment, not
the user-facing production version.
MozReview-Commit-ID: 92HB0pEzeI6
--HG--
extra : rebase_source : ebf00ae704ccd44415f8d7054359c87287734926
GetConsoleWindow() can return NULL. Previously, we may have passed a NULL
console reference into FlashWindowEx(). On my machine (when running in a
console), passing NULL doesn't seem to cause an error. But since we have
a report of this function hanging, it is quite possible it can cause
hangs in other scenarios. Since a NULL console won't result in any
notification, let's not call FlashWindowEx() when no console is available.
MozReview-Commit-ID: LrKX8weUkzX
--HG--
extra : rebase_source : 6c3fc724dbc038f6b51bbdc14d985202c47074e8
The label has been there for years. It isn't really experimental.
The Visual Studio solution still leaves a lot to be desired. But
let's not scare people away by calling it experimental.
MozReview-Commit-ID: 7UvsbsKNnWw
--HG--
extra : rebase_source : a7d3f824859629297a782a0883ebf78af1d8fa70
By using self._write_file() and FileAvoidWrite, we avoid writing files
unless they change. This means that Visual Studio won't want you to
reload the solution and projects whenever the backend is generated.
This means you can regenerate the backend all you want and chances are
it won't disrupt your Visual Studio experience.
Since self._write_file() creates parent directories for us, we were
able to remove this code.
If you run `mach build-backend --diff` with this commit, output will
change. For reasons I don't understand, we were producing XML with
e.g. \r\r\n sequences. This patch appears to restore \r\n. How
we got \r\r I don't know because I can't find anywhere in the code
where that can occur. But this commit does appear to restore sanity.
Also, it appears modern versions of Visual Studio (perhaps only VS2015)
doesn't write your project files. When I initially implemented Visual
Studio project generation several years ago, as soon as you loaded
the solution and hit "Save All" Visual Studio would re-save your files
using a slightly different formatting (it did some gnarly things with
XML indentation). VS2015 doesn't do this. So your files on disk should
be unmodified for longer, making Visual Studio a more viable development
environment. Yay.
MozReview-Commit-ID: 7CSk0dsLOli
--HG--
extra : rebase_source : 3cd04ba8ffed8fc8880de66b1f4fce1f6300b51c
Currently, self._write_file() instantiates FileAvoidWrite instances
with the default mode of 'rU' which uses universal newlines (\n).
Visual Studio project files use CRLF newlines. We want to use
self._write_file() in the Visual Studio backend (which predates
self._write_file). Prepare for this by passing the mode argument
through.
MozReview-Commit-ID: LHCUf3IrpJ8
--HG--
extra : rebase_source : be08d8e043ef625f8f846c38446e1430e072a665
If we're running VS2013, we generate VS2013 files. If we're running VS2015,
we generate VS2015 files. If we don't have a Visual Studio version
defined, refuse to generate project files (hopefully this doesn't
happen in the real world but I'm sure someone will complain if it does).
MozReview-Commit-ID: 5GdsbGmWPLB
--HG--
extra : rebase_source : 1a50c0fa5b7980ad25fb3ed3c1ec2c95f16996b6
We only support building with VS2013 and VS2015. Remove references
to older versions in the Visual Studio build backend.
MozReview-Commit-ID: 6QTSylqLwLF
--HG--
extra : rebase_source : d654f0e415dd5c1bfa1258633e26d467bea5535a
These imports make an out-of-tree build of SpiderMonkey depend on
the reftest module, which means SpiderMonkey implicitly depends on
layout/tools/reftest.
Firefox automation now uploads resource usage JSON files as
job artifacts (see bug 1272202). Now that the build system
writes the same data format and `mach resource-usage` can
read this data format, let's teach `mach resource-usage`
to load arbitrary URLs. This allows people to view system
resource usage for arbitrary jobs in automation.
Currently, you have to look at Treeherder to find the URL to
the build-resources.json artifact. Perhaps in the future
we can make finding the URL easier. Or we could integrate
source resource viewing into Treeherder itself (this is
probably preferred).
This commit continues the tradition of `mach resource-usage`
being a hacked up mess. Instead of loading the URL in
the browser, we download the URL from Python then serve it
from the HTTP server running as part of `mach resource-usage`.
This is somewhat horrible. But it was easiest to implement.
It also conveniently bypasses any cross origin request
restrictions the browser may impose. So it is useful.
MozReview-Commit-ID: IR1Cfs7SrRN
--HG--
extra : rebase_source : b33231d482ca891c1b3331f472009518bf57f8c3
We currently have our own system monitor serialization in
building.py. It predates as_dict() from mozsystemmonitor. Let's
use the "upstream" data format so we only have a single format
to consume.
This change required updating the in-tree resource viewer to
be compatible with the new data format.
This commit stops short of getting rid of the existing
data massaging code in building.py. Another day perhaps.
MozReview-Commit-ID: 1OJrSiyJjMX
--HG--
extra : rebase_source : 9782b2164d1735ed0872fe8c1637204d5b3b1313
Firefox automation now uploads resource usage JSON files as
job artifacts (see bug 1272202). Now that the build system
writes the same data format and `mach resource-usage` can
read this data format, let's teach `mach resource-usage`
to load arbitrary URLs. This allows people to view system
resource usage for arbitrary jobs in automation.
Currently, you have to look at Treeherder to find the URL to
the build-resources.json artifact. Perhaps in the future
we can make finding the URL easier. Or we could integrate
source resource viewing into Treeherder itself (this is
probably preferred).
This commit continues the tradition of `mach resource-usage`
being a hacked up mess. Instead of loading the URL in
the browser, we download the URL from Python then serve it
from the HTTP server running as part of `mach resource-usage`.
This is somewhat horrible. But it was easiest to implement.
It also conveniently bypasses any cross origin request
restrictions the browser may impose. So it is useful.
MozReview-Commit-ID: IR1Cfs7SrRN
--HG--
extra : rebase_source : 91f5f807c19643ac4d1edb8f6652110813f7e53f
We currently have our own system monitor serialization in
building.py. It predates as_dict() from mozsystemmonitor. Let's
use the "upstream" data format so we only have a single format
to consume.
This change required updating the in-tree resource viewer to
be compatible with the new data format.
This commit stops short of getting rid of the existing
data massaging code in building.py. Another day perhaps.
MozReview-Commit-ID: 1OJrSiyJjMX
--HG--
extra : rebase_source : b7c7824b84110f118223dc483b03398855fe9965
Origins will be set for any caller of CommandLineHelper.add, but will only
be propagated if args are added to extra_args. This results in an incorrect
origin recorded for mozconfig injected arguments.
MozReview-Commit-ID: 9mJCaNHyd5C
Originally, the changes to FakeCompiler allowing overlays was meant to
be used for compiler target platform, but it turns out the
simplifications this allows on the compiler definitions themselves are
nice.
We use _pretty_path when specifying the targets of generated files, so
we need to use _pretty_path for the inputs as well. Otherwise make won't
know that they refer to the same file, and result in "No rule to make
target" errors.
MozReview-Commit-ID: JTdLFbkX1J0
Some generated files will depend on other generated files, but still
need to be in the export tier because they are C++ headers.
MozReview-Commit-ID: AFvp92lF0xy
- enable debug artifact from a mozconfig file based on MOZ_DEBUG environment variable
- OSX debug artifact builds have 'mac64' instead of 'mac' into their file name
(fix debug artifact build download on OSX)
MozReview-Commit-ID: 7kAvsTfwaCb
--HG--
extra : transplant_source : %E6v%25%B79M%02%7E%A9%8B%FF%24%03%D1%BDo%AB%0F%B49
Because some older python 2.7 versions throw bogus errors when using the
exec statement as a function, use the function we added in bug 1264831
everywhere we use exec in the various configure tests. It doesn't take
much to trigger them, and the following changes ends up doing exactly
that.
Our current build system support for Rust compiles any Rust crate into a
so-called staticlib, which is a static library (.a file) that includes
the Rust runtime. That staticlib is then linked into libxul. For
supporting multiple crates, this approach breaks down, as linking
multiple copies of the Rust runtime is going to fail.
For supporting multiple crates, the approach taken here is to compile
each crate into a so-called rlib, which is essentially a staticlib
without the Rust runtime linked in. The build system takes note of
every crate destined for linking with libxul (treating them like static
libraries generated from C/C++ files), and generates a super-crate,
whimsically named "rul", that is compiled as a staticlib (so as to
include the Rust runtime) and then linked into libxul. Thus only one
copy of the Rust runtime is included, and the Rust compiler can take
care of any inter-crate dependencies.
This patch currently only supports Rust code in shared libraries, not in
binaries.
The format provided to the build system by the manifest parser is highly redundant:
every test lists all variables for that test, and many tests use a large
support-files entry in DEFAULT that ends up in individual test objects. This
patch stores these DEFAULTS per-manifest rather than per-test to save disk
space, resulting in about a ~22mb smaller all-tests.json file. The
in-memory representation of tests is not changed by this patch, as the defaults
are again propagated to individual tests as all-tests.json is read by the test
resolver.
MozReview-Commit-ID: CEJaevfS5s7
imply_option has no effect when the resolved value is None, so the same
logic can be applied when checking for unknown implied options.
This allows to imply options that may not always exist (because they are
in a configure file that is optionally included).
Ideally, it would be better not to do this, but until we have something
better than optionally included configure files for
--disable-compile-environment, this is a necessary evil.
So far, everything was essentially executed at "declaration". This
made the sandbox code simpler, but to improve on the tooling around
python configure (for tests and introspection), we need to have more
flexibility, which executing everything at declaration doesn't give.
With this change, only @depends functions depending on --help, as
well as templates, are executed at the moment the moz.configure
files are included in the sandbox. The remainder is executed at the
end.
Whether it uses locale._parse_localename or nl_langinfo makes it have completely
different results in weird and/or widespread locale settings (LC_ALL=UTF-8 or
LC_ALL=C).
Check if the INSIDE_EMACS environment variable is set and change the
log level to WARNING to not confuse the emacs/mi with logging messages.
MozReview-Commit-ID: 5AWZ6swGJsE
--HG--
extra : transplant_source : b%24%8Ff6%968%8A%02%E2%07%DD%C6Y9E%CB%7C.%E4
This requires a change to how we process test manifests in the build system:
now, whenever we see a support file mentioned in a manifest, we require that
file isn't already in that test's support files, but if we see a support file
that was already seen in some other test, the entry is ignored, but it is not
an error. As a result of this change, several duplicate support-files entries
needed to be removed.
MozReview-Commit-ID: G0juyxzcaB8
--HG--
rename : testing/mozbase/manifestparser/tests/test_default_skipif.py => testing/mozbase/manifestparser/tests/test_default_overrides.py
Our current build system support for Rust compiles any Rust crate into a
so-called staticlib, which is a static library (.a file) that includes
the Rust runtime. That staticlib is then linked into libxul. For
supporting multiple crates, this approach breaks down, as linking
multiple copies of the Rust runtime is going to fail.
For supporting multiple crates, the approach taken here is to compile
each crate into a so-called rlib, which is essentially a staticlib
without the Rust runtime linked in. The build system takes note of
every crate destined for linking with libxul (treating them like static
libraries generated from C/C++ files), and generates a super-crate,
whimsically named "rul", that is compiled as a staticlib (so as to
include the Rust runtime) and then linked into libxul. Thus only one
copy of the Rust runtime is included, and the Rust compiler can take
care of any inter-crate dependencies.
This patch currently only supports Rust code in shared libraries, not in
binaries. The handling for the rul crate is placed in the common
backend, with a special hook for derived backends to handle shared
library objects.
When reading config.log, with old-configure output, we may get non-ascii
strings, but that currently fails because we're using plain open() to
read it. So use encoded_open() instead (which does the same job for
other files in the same script).
Because the build system can be encapsulated in mach, python configure
can have a pipe as stdout/stderr, and in that case, sys.stdout/stderr
have an ascii encoding, failing to print out anything that doesn't
fit in ascii, consequently failing to print the things we've read from
config.log. So reopen stdout and stderr with the right encoding in
the configure output handler.
This moves test installation for test files out of the monolithic install
manifest for $objdir/_tests, and determines the test and support files
to install based on the object derived from all-tests.json. Additionally,
the files resulting from TEST_HARNESS_FILES are installed, as some tests
will depend on them.
As a result, the time to install tests when invoking the test runner will
scale with the number of tests requested to run rather than the entire set
of tests in the tree, resulting in significantly less overhead.
MozReview-Commit-ID: LeIrUVh1yD4
This extracts the logic from the emitter that handles support files in ini
manifests to a seperate function in testing.py, so that this logic can be
re-used to determine how to install all the files necessary to run a particular
test fon the corresponding object in all-tests.json.
MozReview-Commit-ID: GSEhEGm09IL
DONTBUILD NPOTB
If the target triple is noisy, let's just accept -arm and -i386
instead of being strict. We can be more strict when things have
settled.
MozReview-Commit-ID: FDNJ3TuY51d
--HG--
extra : rebase_source : 3db70abe6818fbe0360406f20de899867232df02
extra : amend_source : d324c229b1713adb844407078d6cc4d00d84b56d
A few notes:
* This doesn't accommodate general OMNIJAR_NAME definitions. The
current name (assets/omni.ja) is baked into the product in a few
places, and is very unlikely to change, so we just error out if this
isn't true.
* This makes the package-manifest.in file authoritative for what goes
into assets/, libs/, and the APK root. Previously,
package-manifest.in wrote into assets/ and libs/ but
upload-files-APK.mk also had a convoluted DIST_FILES filtering
process to work through before a file actually made it into the APK.
* This is intentional about repackaging. It simplifies the repackage
step rather than trying to make unpackage-then-repackage the same as
just package. I pretty much never get repackaging correct the first
time; this should help. (I've manually tested it.)
* The ALREADY_SZIPPED during repackaging is subsumed by the previous
check if UNPACKAGE is set. The custom linker expects stored, not
deflated, libraries, so there's some small legwork to accommodate
that in mozjar.
MozReview-Commit-ID: JvVtIUSX685
--HG--
extra : rebase_source : fd8a9cfe3dc364d23b1065995db599f99e676e38
The initial goal of templates was to provide a way to write shorter
constructs for some generic tasks during configure. But the limitations
of the sandbox and the properties of templates made them used for more
general functions.
Consequently, this led to templates having to be available from
anywhere, which, in turn, led to difficult to introspect constructs.
With bug 1257823, we've made almost everything use set_config and
similar functions from the global scope, but we don't enforce that
those primitives are only used at the global scope.
This change does that: it enforces that primitives are only used at
the global scope. Or in templates.
Now, since templates were used for other purposes than generic uses
of other primitives, we now allow non-template functions to be declared.
Those can be used everywhere, but don't have access to the sandbox
primitives.
Currently, we have @advanced, that gives the decorated functions access
to all the builtins and consequently, to the import statement.
That is a quite broad approach and doesn't allow to easily introspect
what functions are importing which modules.
This change introduces a new decorator that allows to import modules one
by one into the decorated functions.
Note: by the end of the change series, @advanced will be gone.
So far, we've been using the lowercase of the variable name, but it's
not enough for some special cases. Those special cases could do their
own business, but then, they'd have to duplicate 90% of check_prog,
which is less desirable.
While DummyFunction is descriptive of what the instances are (and they
can't even be called), the various uses of isintance(obj, DummyFunction)
are kind of confusing, especially when they are in moz.configure land
(and this bug is about to add another one).
subprocess functions doesn't directly take file-like objects, so add a
minimalistic wrapper to do the right thing instead of subprocess.call
when given a file-like object.
And since the file is also used for old-configure, close our handle on
the file before spawning old-configure, and make old-configure append
there instead of truncating the file.
DMD currently uses a very hacky form of "sampling" by default to avoid
recording stack traces for all blocks. This makes DMD run faster than when it
records all stack traces.
This patch changes the sampling method used; in fact, it avoids "sampling" at
all. The existence of all heap blocks is now recorded exactly, but by default
we only record an allocation stack for each heap block if a Bernoulli trial
succeeds. This choice works well because getting the stack trace is ~100x
slower than recording the block's existence.
Overall, this approach is simpler and it also gives better output -- the choice
of which blocks to record allocation stacks for is mathematically sound, no
stack trace gets blamed for allocations it didn't do, and block counts and
sizes are now always exact.
Other specific things changed.
- All notion of sampling is removed from the various data structures.
- The --sample-below option is removed in favour of --stacks={partial,full}.
- The format of the JSON output file has changed.
- The names of various test files have changed to reflect concept changes.
--HG--
rename : memory/replace/dmd/test/full-empty-cumulative-expected.txt => memory/replace/dmd/test/complete-empty-cumulative-expected.txt
rename : memory/replace/dmd/test/full-empty-dark-matter-expected.txt => memory/replace/dmd/test/complete-empty-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-empty-live-expected.txt => memory/replace/dmd/test/complete-empty-live-expected.txt
rename : memory/replace/dmd/test/full-unsampled1-dark-matter-expected.txt => memory/replace/dmd/test/complete-full1-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-unsampled1-live-expected.txt => memory/replace/dmd/test/complete-full1-live-expected.txt
rename : memory/replace/dmd/test/full-unsampled2-cumulative-expected.txt => memory/replace/dmd/test/complete-full2-cumulative-expected.txt
rename : memory/replace/dmd/test/full-unsampled2-dark-matter-expected.txt => memory/replace/dmd/test/complete-full2-dark-matter-expected.txt
rename : memory/replace/dmd/test/full-sampled-live-expected.txt => memory/replace/dmd/test/complete-partial-live-expected.txt
extra : rebase_source : 47d287405dc5e9075f08addaba49e879c2c6e23f
Before, virtualenv.py may have attempted to use 3rd party
(untrusted) pip indices when installing wheels for pip,
setuptools, and wheel. These dependencies are vendored in
the tree for a reason. So don't let virtualenv contact the
outside world.
MozReview-Commit-ID: 6BCU0WegJO1
--HG--
extra : rebase_source : 6231235486e8cfa784e67e8b18004dd63d8aaf36
@depends functions are declared like the following:
@depends('--option', other_function, '--other-option', ...)
def ...
To simplify some of the processing related to those arguments it's
passed, keep a tuple of Option and DummyFunction objects corresponding
to those arguments.
For the same reasons as set_config is being moved to the global scope,
we're moving set_define to the global scope here. An additional change
is that set_define is now part of the sandbox itself instead of being
defined within the sandbox, which makes it share the implementation
details with set_config.
The way set_config is set currently makes it difficult to introspect
moz.configure files to know what configuration items are being set,
because they're hidden in the control flow of functions.
This makes some of the moz.configure more convoluted, but this is why
there are templates, and we can improve the recurring cases afterwards.
Currently, ConfigureSandbox._db stores two different kind of
information. This split those in two different instance variables
instead, making things clearer.
The way functions are being sandboxed in moz.configure land is that
their global namespace is being replaced with a limited and identifiable
dict. And we avoid re-wrapping a function that already received this
treatment.
The problem is that template functions have their global namespace
replaced, and any function that is defined within the template inherits
that global namespace. So when it comes time to wrap those functions
defined in templates with e.g. depends, we detect that they're already
wrapped although they are not, because we look if their global namespace
is of the recognizable type we use when replacing it.
So instead of looking at the global namespace type, keep track of all
functions that are wrapped.
Currently, if a @depends function doesn't have a return statement or
return None, a result is automatically set from all the set_config()
called from the function.
As we're going to move set_config to the global namespace, and as this
feature is only used once, and it's only used for something that was
written before ReadOnlyNamespace was exposed to the sandbox, we can
"safely" get rid of it.
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
This change adds a `Version` type to moz.configure which is a small
wrapper around `distutils.version.Version`. It's suitable for wrapping
version numbers in configure checks and doing equality or greater-than
less-than comparisons in a sensible way.
MozReview-Commit-ID: BOL6yvemulG
--HG--
extra : rebase_source : 3b463eac0499086f8acffda0d01418b6ab17f3d6
extra : amend_source : aebd6e40c408d9f868623b2f53fcdf7455e2fff5
In Python parens around an expression without a trailing comma is not a tuple,
so ('foo') == 'foo'. This is really easy to screw up with check_progs, which
coerces progs to a list and would give you ['f','o','o'] in this case. This
patch enforces that the progs argument is a tuple or list and errors if it
is not.
MozReview-Commit-ID: 7BJZuF9B8D5
--HG--
extra : rebase_source : 5db9a6b4bb9ef7195c2513407810093bff5e9174
extra : amend_source : f67ab46c2ac00a2a95cfc67e9763ac12b690ac14
If we add an ObjDirPath as an input to a GENERATED_FILES script, we
should run it in the misc tier to ensure the dependent files are created
beforehand. We need to use Path objects instead of raw filenames in the
GeneratedFile object so the recursive make backend can distinguish
between source and objdir files.
MozReview-Commit-ID: 9thHTi75zdI
While rare, this is something we support for e.g. --enable-eme, where it
can be either given with no values, "adobe", some future other EME GMP
adapter names, or a combination of them.
This will make it easier for binaries only archive generation to consult
the list of binaries that are relevant to packaging.
This does overlap somewhat with compile databases. Perhaps someday the
logic could converge.
While I was here, I cleaned up writing of the all-tests.json file to
avoid an intermediate string variable.
This patch adds to_dict methods on some frontend data types. There is
room to improve the serialization of these types. For example, we could
use __slots__ to drive the default formatter. We could also leverage a
custom JSONEncoder class that knows how to call to_dict() on instances.
In the spirit of perfect is the enemy of done, I think this should be
deferred to a follow-up bug.
MozReview-Commit-ID: XVnKB1MNqu
--HG--
extra : source : 4167dfdf10457360c9c94ee6e55b03ef1b92c16d
extra : histedit_source : 2f3f10bbb8ef9da8c57b0d85cbf4ea80242eff1a%2C1a5654253181d75eecd5dfaca2c87e353f8126cd
Now when we print instances in a debugger, we'll see something like:
<StaticLibrary: other-licenses/snappy/libother-licenses_snappy.a>
<StaticLibrary: toolkit/library/StaticXULComponentsEnd/libStaticXULComponentsEnd.a>
<SharedLibrary: toolkit/library/XUL>
<StaticLibrary: toolkit/library/libxul_s.a>
<HostProgram: config/nsinstall_real>
<Program: memory/replace/logalloc/replay/logalloc-replay>
<SimpleProgram: mfbt/tests/TestArrayUtils>
... instead of the the class name and memory address.
MozReview-Commit-ID: 8zdrM6KfP8U
--HG--
extra : source : 84849ad026c9ba1bbf71c93172b0a03440e51bec
This will make it easier for binaries only archive generation to consult
the list of binaries that are relevant to packaging.
This does overlap somewhat with compile databases. Perhaps someday the
logic could converge.
While I was here, I cleaned up writing of the all-tests.json file to
avoid an intermediate string variable.
This patch adds to_dict methods on some frontend data types. There is
room to improve the serialization of these types. For example, we could
use __slots__ to drive the default formatter. We could also leverage a
custom JSONEncoder class that knows how to call to_dict() on instances.
In the spirit of perfect is the enemy of done, I think this should be
deferred to a follow-up bug.
MozReview-Commit-ID: XVnKB1MNqu
--HG--
extra : rebase_source : 376c93467dde3450c7c21cff918cb34913a3c5fe
Now when we print instances in a debugger, we'll see something like:
<StaticLibrary: other-licenses/snappy/libother-licenses_snappy.a>
<StaticLibrary: toolkit/library/StaticXULComponentsEnd/libStaticXULComponentsEnd.a>
<SharedLibrary: toolkit/library/XUL>
<StaticLibrary: toolkit/library/libxul_s.a>
<HostProgram: config/nsinstall_real>
<Program: memory/replace/logalloc/replay/logalloc-replay>
<SimpleProgram: mfbt/tests/TestArrayUtils>
... instead of the the class name and memory address.
MozReview-Commit-ID: 8zdrM6KfP8U
--HG--
extra : rebase_source : 7f964ad44a0061674f77d5716a6769a4aedb9e6c
I'm not sure when this changed, but at least Visual Studio 2015
doesn't always emit a space between the line number and the ": warning"
text in cl.exe output.
Making the space optional in the regular expression enables one a
VS2015 build to capture 375 warnings instead of 17. We still fail to
capture some warnings (notably generic warnings about bad command
arguments and linker warnings). But that can be dealt with later.
MozReview-Commit-ID: q402CxTrQK
--HG--
extra : rebase_source : 6376a65b8d8145d68bad9b795b6abd6f4becefbd
Virtualenv will sometimes find a different executable from its sys.executable on OS X,
causing a check in the build system comparing filesizes between sys.executable and virtualenv
python to fail, resulting in clobbering and re-building the virtualenv every time the virtualenv
is activated, causing the build backend and more to be re-built.
Instead of checking file sizes directly, this commit causes us to record the size and version of the
Python executable that created the virtualenv. If the Python executable checked is not the virtualenv
Python, or we have a different version than was used to create the virtualenv, then the virtualenv is
considered to be out of date.
MozReview-Commit-ID: KmrVfQCtbS3
--HG--
extra : rebase_source : d2b87325e10da6dfcd74f8b0d2ef7c0efb71595b
Virtualenv will sometimes find a different executable from its sys.executable on OS X,
causing a check in the build system comparing filesizes between sys.executable and virtualenv
python to fail, resulting in clobbering and re-building the virtualenv every time the virtualenv
is activated, causing the build backend and more to be re-built.
Instead of checking file sizes directly, this commit causes us to record the size and path to the Python
executable that created the virtualenv. If the Python executable checked is not the virtualenv
Python, or the python that was used to create the interpreter, then the virtualenv is considered
to be out of date.
MozReview-Commit-ID: KmrVfQCtbS3
When writing e.g. Japanese text files on Windows, the text is encoded in
mbcs. So if a mozconfig is edited this way, and contains non ascii
characters, they won't be utf-8 as we're currently trying to decode
them, but mbcs.
This adds support for an SDK_FILES variable in moz.build, which creates
a FinalTargetFiles object to install files into dist/sdk/
MozReview-Commit-ID: 97a5NdbZmmD
--HG--
extra : rebase_source : ad8d521ec56fe4610437c8d2d503c545894b40c2
The changes made to PATH by activating the in-tree virtualenv cause some git-cinnabar commands
in subprocesses to fail, because git cinnabar is expecting mercurial to be installed, while it
is not available from the in-tree virtualenv. To work around this, the changes to PATH made by
virtualenv are undone, allowing git-cinnabar to find system python, which is likely to have
mercurial installed.
MozReview-Commit-ID: 47ceHvMmuyf
--HG--
extra : rebase_source : 199d3b01fc49b2760ee14a08d4400a1addf83eac
This moves all the reading mozconfig, finding autoconf, refreshing the
old configure, and running the old configure into sandboxed
moz.configure. This effectively bootstraps the sandboxed python configure.
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.
The upcoming move of the configure.py initialization to sandboxed
moz.configure changes the path separators for topsrcdir and topobjdir
from native to always use forward-slashes, which confuses the hell out
of the test_base.py test. Settle the issue by declaring that
MozbuildObject will always use forward-slashed paths for topsrcdir
and topobjdir.
The nice side effect is that now we can have actual dicts for defines
and substs from the start, which simplifies so things, although it
requires adjustments to some unit tests.
Mike Shal
Al Tsai
Mike Hommey
Henrik Skupin
Xidorn Quan
Myk Melez
Gregory Szorc
Ryan VanderMeulen
Wes Kocher
Chris Manchester
Elias Probst
Carsten "Tomcat" Book
Nathan Froyd
Ehsan Akhgari
Nick Alexander
Geoff Brown
Tom Tromey
Xidorn Quan
Nathan Hakkakzadeh
Andrew Halberstadt
Gijs Kruitbosch
Ted Mielczarek
Justin D'Arcangelo
Chris AtLee
Ms2ger
Timothy Guan-tin Chien
Stephen Pohl
Mark Hammond
Ben Kelly
Luca Greco
Matt Howell
Byron Campen [:bwc]
Nathan Froyd
Chris Pearce
Andreas Farre
Sebastian Hengst
George Wright
Nicholas Nethercote
aleth
Richard Newman
Phil Ringnalda
Maja Frydrychowicz