1319222 - Support compilation objects in the tup backend

Status: RESOLVED FIXED

Description

[Michael Shal [:mshal]]

We'll need to compile all of the *Sources moz.build variables in the tup backend. I plan to share most of the CompilationDB command-line generation logic in the CommonBackend, and use that for the tup backend.

Comment 1

[Michael Shal [:mshal]]

Created attachment 8931140 [details]
Bug 1319222 - Remove srcdir from BackendTupfile;

The srcdir does not necessarily directly correspond to the objdir, so it
doesn't make sense to tie them together in BackendTupfile. Since the
srcdir was only used for the IPDL sources, we can just replace that
usage with a local variable.

Review commit: https://reviewboard.mozilla.org/r/202222/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202222/

Comment 2

[Michael Shal [:mshal]]

Created attachment 8931141 [details]
Bug 1319222 - Use relobjdir instead of relativedir to determine outputs;

In most cases, relobjdir is the same as relativedir. However they are
different for some objects, notably with nss gyp handling.  We need to
output files in relobjdir, so use that when getting the BackendTupfile.
This puts generated files like certdata.c in the directory that
moz.build is expecting.

Review commit: https://reviewboard.mozilla.org/r/202224/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202224/

Comment 3

[Michael Shal [:mshal]]

Created attachment 8931142 [details]
Bug 1319222 - Change --param flag to be multiple arguments;

This flag should be two parameters, the first of which is '--param', and
the second is the parameter, such as 'inline-unit-growth=10000'. The
recursive make backend lets this slide at the moment because
PerSourceFlags aren't shell_quoted() there.

Review commit: https://reviewboard.mozilla.org/r/202226/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202226/

Comment 4

[Michael Shal [:mshal]]

Created attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

These flags are all needed to get the full command-line for assembly
compilation. An example of a file that fails without them is unix64.S in
objdir/config/external/ffi.

Review commit: https://reviewboard.mozilla.org/r/202228/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202228/

Comment 5

[Michael Shal [:mshal]]

Created attachment 8931144 [details]
Bug 1319222 - Compile the generated IPDL and WebIDL sources in the tup backend;

The filenames that these objects generate are passed into the _handle_*
methods instead of with a Sources object, so they need to be added to
the BackendTupfile's list of sources separately.

Review commit: https://reviewboard.mozilla.org/r/202230/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202230/

Comment 6

[Michael Shal [:mshal]]

Created attachment 8931145 [details]
Bug 1319222 - Enable compilation on all directories in the tup backend;

Review commit: https://reviewboard.mozilla.org/r/202232/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202232/

Comment 7

[Michael Shal [:mshal]]

Created attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

This accounts for LOCAL_INCLUDES in the moz.build files, as well as the
default INCLUDES specified in config.mk that are used for host
compilation.

Review commit: https://reviewboard.mozilla.org/r/202234/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202234/

Comment 8

[Michael Shal [:mshal]]

Created attachment 8931147 [details]
Bug 1319222 - Enable host compilation in the tup backend;

Review commit: https://reviewboard.mozilla.org/r/202236/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202236/

Comment 9

[Michael Shal [:mshal]]

Created attachment 8931148 [details]
Bug 1319222 - Support compiling *.s with yasm;

The moz.build files that specify USE_YASM = True will get the value of
AS_DASH_C_FLAG and AS overwritten in PassthruVariables. We can save
those in the BackendTupfile and use them in place of the configure or
default settings as appropriate. This enables compilation of those .s
files that are built with yasm.

Review commit: https://reviewboard.mozilla.org/r/202238/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/202238/

Comment 10

[Michael Shal [:mshal]]

With this set of patches, a build with the tup backend build will generate the same set of files as a recursive make build with an empty mozconfig. I haven't yet verified that the object files are identical, but progress! And actually linking them will be handled in other bugs.

I requested chmanchester for the flags patches since I'm really not sure if those are implemented correctly. They seemed to be required for host and asm compilation in a few places.

Comment 11

[Chris Manchester (:chmanchester)]

Comment on attachment 8931140 [details]
Bug 1319222 - Remove srcdir from BackendTupfile;

https://reviewboard.mozilla.org/r/202222/#review207582

Comment 12

[Chris Manchester (:chmanchester)]

Comment on attachment 8931141 [details]
Bug 1319222 - Use relobjdir instead of relativedir to determine outputs;

https://reviewboard.mozilla.org/r/202224/#review207584

::: python/mozbuild/mozbuild/backend/tup.py:207
(Diff revision 1)
>                      BackendTupfile(objdir, self.environment,
>                                     self.environment.topsrcdir, self.environment.topobjdir)
>          return self._backend_files[objdir]
>  
>      def _get_backend_file_for(self, obj):
> -        return self._get_backend_file(obj.relativedir)
> +        return self._get_backend_file(obj.relobjdir)

Wow, this is sort of a footgun. I would say let's re-name `relativedir` to `relsrcdir`... in the constructor for ContextDerived we actually do `self.relativedir = context.relsrcdir`. If you'd rather not fix that here let's do it in a follow up.

I guess this explains all the times I've failed trying to find `relsrcdir` on an object in the build system.

Comment 13

[Chris Manchester (:chmanchester)]

Comment on attachment 8931142 [details]
Bug 1319222 - Change --param flag to be multiple arguments;

https://reviewboard.mozilla.org/r/202226/#review207586

Comment 14

[Chris Manchester (:chmanchester)]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

https://reviewboard.mozilla.org/r/202228/#review207590

::: python/mozbuild/mozbuild/frontend/context.py:353
(Diff revision 1)
> +            ('DEFINES', None, ('ASFLAGS',)),
> +            ('LIBRARY_DEFINES', None, ('ASFLAGS',)),
> +            ('LOCAL_INCLUDES', None, ('ASFLAGS',)),

There are two potential issues here:

DEFINES and INCLUDES are added to the command line with ASFLAGS for SOBJS (made from `*.S` files), but not for ASOBJS (made from `*.s` files). Also, by adding them here but not deleting them from the rules.mk command line we'd end up with them duplicated in the Make backend. Probably we can split these off into a separate variable if this distinction is really significant.

The second potential issue is that in the Make backend DEFINES come before ASFLAGS, followed by per-source flags, followed by INCLUDES. Changing the order could be significant. I doubt it is, but we would want to check. If we want to preserve the order we could probably just cobble this together in the backend.

::: python/mozbuild/mozbuild/frontend/emitter.py:1237
(Diff revision 1)
>          if passthru.variables:
>              yield passthru
>  
>          if context.objdir in self._compile_dirs:
>              self._compile_flags[context.objdir] = computed_flags
> +            self._compile_as_flags[context.objdir] = computed_as_flags

This would make `_asm_compile_dirs` obsolete. Was that too restrictive for some reason?

Comment 15

[Chris Manchester (:chmanchester)]

Comment on attachment 8931144 [details]
Bug 1319222 - Compile the generated IPDL and WebIDL sources in the tup backend;

https://reviewboard.mozilla.org/r/202230/#review207592

::: python/mozbuild/mozbuild/backend/tup.py:569
(Diff revision 1)
>              cmd=cmd,
>              outputs=outputs,
>              extra_outputs=[self._installed_files],
>              check_unchanged=True,
>          )
> +        backend_file.sources['.cpp'].extend([u[0] for u in unified_ipdl_cppsrcs_mapping])

Intermediate list not needed?

::: python/mozbuild/mozbuild/backend/tup.py:602
(Diff revision 1)
>              inputs=webidls.all_non_static_basenames(),
>              outputs=outputs,
>              extra_outputs=[self._installed_files],
>              check_unchanged=True,
>          )
> +        backend_file.sources['.cpp'].extend([u[0] for u in unified_source_mapping])

Intermediate list not needed?

Comment 16

[Chris Manchester (:chmanchester)]

Comment on attachment 8931145 [details]
Bug 1319222 - Enable compilation on all directories in the tup backend;

https://reviewboard.mozilla.org/r/202232/#review207594

Comment 17

[Chris Manchester (:chmanchester)]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

https://reviewboard.mozilla.org/r/202234/#review207600

::: python/mozbuild/mozbuild/frontend/context.py:332
(Diff revision 1)
> +            ('BASE_INCLUDES', ['-I%s' % main_src_dir, '-I%s' % context.objdir],
> +             ('HOST_CFLAGS', 'HOST_CXXFLAGS')),
> +            ('LOCAL_INCLUDES', None, ('HOST_CFLAGS', 'HOST_CXXFLAGS')),
> +            ('EXTRA_INCLUDES', ['-I%s/dist/include' % context.config.topobjdir],
> +             ('HOST_CFLAGS', 'HOST_CXXFLAGS')),

This has a sort of similar issue to `ASFLAGS`: the includes aren't passed every time we see `HOST_CFLAGS`, and we'd have to delete these from `rules.mk` if we were to add them here. Again we probably need to shunt these off to a separate variable or just stitch things together in the tup backend for now.

Comment 18

[Chris Manchester (:chmanchester)]

Comment on attachment 8931147 [details]
Bug 1319222 - Enable host compilation in the tup backend;

https://reviewboard.mozilla.org/r/202236/#review207602

::: python/mozbuild/mozbuild/backend/tup.py:122
(Diff revision 1)
> +        s = self.sources
> +        h = self.host_sources

Nit: one char variables are hard to grep for.

::: python/mozbuild/mozbuild/backend/tup.py:125
(Diff revision 1)
> -            ('.S', 'AS', 'ASFLAGS'),
> -            ('.cpp', 'CXX', 'CXXFLAGS'),
> -            ('.c', 'CC', 'CFLAGS'),
> +            (s['.S'], 'AS', 'ASFLAGS', ''),
> +            (s['.cpp'], 'CXX', 'CXXFLAGS', ''),
> +            (s['.c'], 'CC', 'CFLAGS', ''),
> +            (h['.cpp'], 'HOST_CXX', 'HOST_CXXFLAGS', 'host_'),
> +            (h['.c'], 'HOST_CC', 'HOST_CFLAGS', 'host_'),

I just noticed the recursivemake backend cares about the difference between `.s` and `.S`. Sorry, I probably should have caught this is a prior review :/

Comment 19

[Chris Manchester (:chmanchester)]

Comment on attachment 8931147 [details]
Bug 1319222 - Enable host compilation in the tup backend;

https://reviewboard.mozilla.org/r/202236/#review207602

> I just noticed the recursivemake backend cares about the difference between `.s` and `.S`. Sorry, I probably should have caught this is a prior review :/

Oh, you take care of this in the next patch. Nevermind!

Comment 20

[Chris Manchester (:chmanchester)]

Comment on attachment 8931148 [details]
Bug 1319222 - Support compiling *.s with yasm;

https://reviewboard.mozilla.org/r/202238/#review207606

::: python/mozbuild/mozbuild/backend/tup.py:142
(Diff revision 1)
> -                cmd.extend(['-c', '%f', '-o', '%o'])
> +                if compiler == 'AS':
> +                    cmd.append(self.variables.get('AS_DASH_C_FLAG', '-c'))
> +                else:
> +                    cmd.append('-c')

Can this whole thing just be `if compile_value != self.environment.substs['YASM']: cmd.append('-c')`? 

That way we can avoid a reference to `AS_DASH_C_FLAG`.

Comment 21

[Michael Shal [:mshal]]

Comment on attachment 8931141 [details]
Bug 1319222 - Use relobjdir instead of relativedir to determine outputs;

https://reviewboard.mozilla.org/r/202224/#review207584

> Wow, this is sort of a footgun. I would say let's re-name `relativedir` to `relsrcdir`... in the constructor for ContextDerived we actually do `self.relativedir = context.relsrcdir`. If you'd rather not fix that here let's do it in a follow up.
> 
> I guess this explains all the times I've failed trying to find `relsrcdir` on an object in the build system.

Filed bug 1421038 as a followup.

Comment 22

[Michael Shal [:mshal]]

Comment on attachment 8931148 [details]
Bug 1319222 - Support compiling *.s with yasm;

https://reviewboard.mozilla.org/r/202238/#review207606

> Can this whole thing just be `if compile_value != self.environment.substs['YASM']: cmd.append('-c')`? 
> 
> That way we can avoid a reference to `AS_DASH_C_FLAG`.

That does seem to work, though that makes this logic a little redundant with how AS_DASH_C_FLAG gets set in the emitter. I think it might be confusing to have that in the emitter and go unused in the tup backend.

I've re-worked it slightly to specify the dash_c setting in the compilers list so it no longer uses the weird if-statement. I'll ask for re-review on this one.

Comment 23

[Michael Shal [:mshal]]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

https://reviewboard.mozilla.org/r/202228/#review207590

> There are two potential issues here:
> 
> DEFINES and INCLUDES are added to the command line with ASFLAGS for SOBJS (made from `*.S` files), but not for ASOBJS (made from `*.s` files). Also, by adding them here but not deleting them from the rules.mk command line we'd end up with them duplicated in the Make backend. Probably we can split these off into a separate variable if this distinction is really significant.
> 
> The second potential issue is that in the Make backend DEFINES come before ASFLAGS, followed by per-source flags, followed by INCLUDES. Changing the order could be significant. I doubt it is, but we would want to check. If we want to preserve the order we could probably just cobble this together in the backend.

Ahh, good catch. I agree it's probably easiest to separate the flags for SOBJS out into a separate variable. I've re-worked this commit to go with that approach, and verified locally that the before & after commandlines in the make backend are identical for a .s and a .S file.

> This would make `_asm_compile_dirs` obsolete. Was that too restrictive for some reason?

I think this issue is no longer relevant with the new approach, since the original AsmFlags are the same as before. Let me know if I'm misunderstanding.

Comment 24

[Michael Shal [:mshal]]

Comment on attachment 8931140 [details]
Bug 1319222 - Remove srcdir from BackendTupfile;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202222/diff/1-2/

Comment 25

[Michael Shal [:mshal]]

Comment on attachment 8931141 [details]
Bug 1319222 - Use relobjdir instead of relativedir to determine outputs;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202224/diff/1-2/

Comment 26

[Michael Shal [:mshal]]

Comment on attachment 8931142 [details]
Bug 1319222 - Change --param flag to be multiple arguments;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202226/diff/1-2/

Comment 27

[Michael Shal [:mshal]]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202228/diff/1-2/

Comment 28

[Michael Shal [:mshal]]

Comment on attachment 8931144 [details]
Bug 1319222 - Compile the generated IPDL and WebIDL sources in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202230/diff/1-2/

Comment 29

[Michael Shal [:mshal]]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202234/diff/1-2/

Comment 30

[Michael Shal [:mshal]]

Comment on attachment 8931147 [details]
Bug 1319222 - Enable host compilation in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202236/diff/1-2/

Comment 31

[Michael Shal [:mshal]]

Comment on attachment 8931148 [details]
Bug 1319222 - Support compiling *.s with yasm;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202238/diff/1-2/

Comment 32

[Michael Shal [:mshal]]

Comment on attachment 8931145 [details]
Bug 1319222 - Enable compilation on all directories in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202232/diff/1-2/

Comment 33

[Chris Manchester (:chmanchester)]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

https://reviewboard.mozilla.org/r/202228/#review209980

Looks good, thanks for doing this!

::: python/mozbuild/mozbuild/frontend/context.py:2034
(Diff revision 2)
> -        """Recipe for linker flags for this context. Not to be manipulated
> -        directly.
> +        """Recipe for assembler flags for this context, without define or
> +        include flags since they are not supported by some assemblers.  Not to

This comment isn't quite right now (I don't think it needs to be updated). The `ASM_FLAGS` dict will have everything but decide which flags categories to pass on to `ASFLAGS` vs `SFLAGS` based on the `flag_variables` definition above.

::: python/mozbuild/mozbuild/frontend/emitter.py
(Diff revision 2)
>              passthru.variables['AS'] = yasm
>              passthru.variables['AS_DASH_C_FLAG'] = ''
>              computed_as_flags.resolve_flags('OS',
>                                              context.config.substs.get('YASM_ASFLAGS', []))
>  
> -

Stray line deletion.

Comment 34

[Chris Manchester (:chmanchester)]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

https://reviewboard.mozilla.org/r/202234/#review209990

::: config/config.mk:212
(Diff revision 2)
> +HOST_LINK_CFLAGS = $(COMPUTED_HOST_LINK_CFLAGS)
> +HOST_LINK_CXXFLAGS = $(COMPUTED_HOST_LINK_CXXFLAGS)

For `CompileFlags` we have `C{XX}_LDFLAGS` for the ldflags we also pass to the linked. I guess I'll say let's keep that naming convention, although it's not a perfect fit because for whatever reason we pass includes to link `HOST_SIMPLE_PROGRAMS`. Let's just add a comment or something to the commit message acknowledging the difference between the `HOST_SIMPLE_PROGRAM` and `HOST_PROGRAM` command lines.

Comment 35

[Michael Shal [:mshal]]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

https://reviewboard.mozilla.org/r/202228/#review209980

> Stray line deletion.

Whoops, fixed!

Comment 36

[Michael Shal [:mshal]]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

https://reviewboard.mozilla.org/r/202234/#review209990

> For `CompileFlags` we have `C{XX}_LDFLAGS` for the ldflags we also pass to the linked. I guess I'll say let's keep that naming convention, although it's not a perfect fit because for whatever reason we pass includes to link `HOST_SIMPLE_PROGRAMS`. Let's just add a comment or something to the commit message acknowledging the difference between the `HOST_SIMPLE_PROGRAM` and `HOST_PROGRAM` command lines.

Ok, the name is updated.

As for the differences in flag usage between HOST_SIMPLE_PROGRAM and HOST_PROGRAM, it doesn't look like we need INCLUDES or DEFINES in the linker rule for HOST_SIMPLE_PROGRAM anymore (it's only used in layout/style/test for host_ListCSSProperties, and there is a separate compiler step anyway). How about just changing the HOST_SIMPLE_PROGRAM flags to use HOST_CXX_LDFLAGS instead of HOST_CXXFLAGS?

Comment 37

[Chris Manchester (:chmanchester)]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

https://reviewboard.mozilla.org/r/202234/#review209990

> Ok, the name is updated.
> 
> As for the differences in flag usage between HOST_SIMPLE_PROGRAM and HOST_PROGRAM, it doesn't look like we need INCLUDES or DEFINES in the linker rule for HOST_SIMPLE_PROGRAM anymore (it's only used in layout/style/test for host_ListCSSProperties, and there is a separate compiler step anyway). How about just changing the HOST_SIMPLE_PROGRAM flags to use HOST_CXX_LDFLAGS instead of HOST_CXXFLAGS?

By all means, if the difference isn't needed anymore let's get rid of it! Let's add this as a "pre" patch.

Comment 38

[Chris Manchester (:chmanchester)]

Replied in mozreview.

Comment 39

[Michael Shal [:mshal]]

Created attachment 8934596 [details]
Bug 1319222 - Remove INCLUDES from HOST_SIMPLE_PROGRAMS linking;

These flags were added way back in bug 59454 to both the compiling and
linking rules for HOST_SIMPLE_PROGRAMS. However, the INCLUDES aren't
actually needed when linking so we can safely remove them.

Review commit: https://reviewboard.mozilla.org/r/205482/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/205482/

Comment 40

[Michael Shal [:mshal]]

Comment on attachment 8931140 [details]
Bug 1319222 - Remove srcdir from BackendTupfile;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202222/diff/2-3/

Comment 41

[Michael Shal [:mshal]]

Comment on attachment 8931141 [details]
Bug 1319222 - Use relobjdir instead of relativedir to determine outputs;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202224/diff/2-3/

Comment 42

[Michael Shal [:mshal]]

Comment on attachment 8931142 [details]
Bug 1319222 - Change --param flag to be multiple arguments;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202226/diff/2-3/

Comment 43

[Michael Shal [:mshal]]

Comment on attachment 8931143 [details]
Bug 1319222 - Add an SFLAGS ComputedFlags variable for compiling *.S;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202228/diff/2-3/

Comment 44

[Michael Shal [:mshal]]

Comment on attachment 8931144 [details]
Bug 1319222 - Compile the generated IPDL and WebIDL sources in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202230/diff/2-3/

Comment 45

[Michael Shal [:mshal]]

Comment on attachment 8931146 [details]
Bug 1319222 - Add include flags to HOST_*;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202234/diff/2-3/

Comment 46

[Michael Shal [:mshal]]

Comment on attachment 8931147 [details]
Bug 1319222 - Enable host compilation in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202236/diff/2-3/

Comment 47

[Michael Shal [:mshal]]

Comment on attachment 8931148 [details]
Bug 1319222 - Support compiling *.s with yasm;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202238/diff/2-3/

Comment 48

[Michael Shal [:mshal]]

Comment on attachment 8931145 [details]
Bug 1319222 - Enable compilation on all directories in the tup backend;

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/202232/diff/2-3/

Comment 49

[Chris Manchester (:chmanchester)]

Comment on attachment 8934596 [details]
Bug 1319222 - Remove INCLUDES from HOST_SIMPLE_PROGRAMS linking;

https://reviewboard.mozilla.org/r/205482/#review211140

Comment 50

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/05b9eb8ae66b
Remove srcdir from BackendTupfile; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/1c5d8c7e5ddd
Use relobjdir instead of relativedir to determine outputs; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/598cfad59614
Change --param flag to be multiple arguments; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/8e89b51b7f40
Add an SFLAGS ComputedFlags variable for compiling *.S; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/a35541618d92
Compile the generated IPDL and WebIDL sources in the tup backend; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/d115dc76ef73
Remove INCLUDES from HOST_SIMPLE_PROGRAMS linking; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/8b0115b2fed8
Add include flags to HOST_*; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/b6d367955ce6
Enable host compilation in the tup backend; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/7f12adb5e683
Support compiling *.s with yasm; r=chmanchester
https://hg.mozilla.org/integration/autoland/rev/a3160ab7a6bf
Enable compilation on all directories in the tup backend; r=chmanchester

Comment 51

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/05b9eb8ae66b
https://hg.mozilla.org/mozilla-central/rev/1c5d8c7e5ddd
https://hg.mozilla.org/mozilla-central/rev/598cfad59614
https://hg.mozilla.org/mozilla-central/rev/8e89b51b7f40
https://hg.mozilla.org/mozilla-central/rev/a35541618d92
https://hg.mozilla.org/mozilla-central/rev/d115dc76ef73
https://hg.mozilla.org/mozilla-central/rev/8b0115b2fed8
https://hg.mozilla.org/mozilla-central/rev/b6d367955ce6
https://hg.mozilla.org/mozilla-central/rev/7f12adb5e683
https://hg.mozilla.org/mozilla-central/rev/a3160ab7a6bf