Don't filter by extensions.

Don't filter by extensions.

When using the -all option in run_tool.py, don't filter the files
in the compilation database by extension. Otherwise, files ending
in ".S", ".cxx" and ".cpp" are filtered out.

BUG=429684
Committed: https://crrev.com/d2db601496fc7ccbf0f689bbcafa3796e7c15141
Cr-Commit-Position: refs/heads/master@{#305445}

[Adrian Kuegel, 2014-11-20 12:41:47]

akuegel@chromium.org changed reviewers:
+ dcheng@chromium.org

[Adrian Kuegel, 2014-11-20 12:41:48]

Hi Daniel,

can you please review this CL? I was missing that change in my CL yesterday.
Maybe it also makes sense to update the extensions which are processed? So as
far as I have seen, the only files missing were the ones ending in ".cxx", ".S"
and ".cpp".

[Adrian Kuegel, 2014-11-24 16:57:23]

On 2014/11/20 12:41:48, Adrian Kuegel wrote:
> Hi Daniel,
> 
> can you please review this CL? I was missing that change in my CL yesterday.
> Maybe it also makes sense to update the extensions which are processed? So as
> far as I have seen, the only files missing were the ones ending in ".cxx",
".S"
> and ".cpp".

Daniel, can you please take a look?

[dcheng, 2014-11-24 17:03:20]

It would be nice to not repeat the _CompilerDispatcher invocation. Maybe
filenames shouldn't be a frozenset?

[Adrian Kuegel, 2014-11-24 17:14:17]

On 2014/11/24 17:03:20, dcheng wrote:
> It would be nice to not repeat the _CompilerDispatcher invocation. Maybe
> filenames shouldn't be a frozenset?

I tried to avoid it. I am not sure if the conversion to set() would give any
speedup here, or if it could be avoided as well? Or did you already mean that we
don't need a set at all?

[dcheng, 2014-11-24 17:23:45]

I used a set originally to dedupe filenames. It's up to you if you think it's
important to dedupe for compile databases as well (I think it is actually
possible for there to be duplicate files in there, due to the way compile DB
entries are emitted for stuff that gets compiled into non-NaCl and NaCl code).

LGTM either way.

[Adrian Kuegel, 2014-11-24 17:28:33]

On 2014/11/24 17:23:45, dcheng wrote:
> I used a set originally to dedupe filenames. It's up to you if you think it's
> important to dedupe for compile databases as well (I think it is actually
> possible for there to be duplicate files in there, due to the way compile DB
> entries are emitted for stuff that gets compiled into non-NaCl and NaCl code).
> 
> LGTM either way.

You are right, there are duplicates. I had forgot about those. I now also use a
set for the "--all" option.

[Adrian Kuegel, 2014-11-24 17:28:45]

The CQ bit was checked by akuegel@chromium.org

[commit-bot: I haz the power, 2014-11-24 17:29:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/743133002/40001

[commit-bot: I haz the power, 2014-11-24 18:10:25]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2014-11-24 18:11:09]

Patchset 3 (id:??) landed as
https://crrev.com/d2db601496fc7ccbf0f689bbcafa3796e7c15141
Cr-Commit-Position: refs/heads/master@{#305445}