It all builds on Linux!

[Hans Wennborg]

Since llvm.org/pr3933 was fixed, it's now possible to "make all" of
Chromium on Linux using Clang. Hooray! :)

(I used Clang built from r122869.)

I think we can now update http://code.google.com/p/chromium/wiki/Clang
to remove the notes about setting linux_use_tcmalloc=0 ?

And I suppose we should also update the Linux bot to make all, and run
some more tests?

I ran a bunch of the tests (these were all compiled in release mode
with Clang, Chromium r70491):

base_unittests: fails ProcessUtilTest.GetTerminationStatusCrash
googleurl_unittests: ok
gpu_unittests: ok
media_unittests: ok
net_unittests: ok
printing_unittests: ok
remoting_unittests: ok
test_shell_tests: fails ICOImageDecoderTest.Decoding, and a few more
ipc_tests: ok
sync_unit_tests: ok
unit_tests: fails DownloadUtilTest.GenerateFileName (don't think it's
clang's fault)
app_unittests: ok
browser_tests: failed ExtensionApiTest.BookmarkManager
ui_tests: ok
webkit_unit_tests: ok
memory_test: ok

I'll try to find time to look into the failures, especially
test_shell_tests sounds interesting.

Cheers,
Hans

[Evan Martin]

To track work on all of this, I suggest filing bugs with an additional
"Clang" label.

On Wed, Jan 5, 2011 at 5:47 AM, Hans Wennborg <hwen...@google.com> wrote:
> I think we can now update http://code.google.com/p/chromium/wiki/Clang
> to remove the notes about setting linux_use_tcmalloc=0 ?

bug 1

> And I suppose we should also update the Linux bot to make all, and run
> some more tests?

bug 2

> base_unittests: fails ProcessUtilTest.GetTerminationStatusCrash

bug 3

> test_shell_tests: fails ICOImageDecoderTest.Decoding, and a few more

bug 4

> unit_tests: fails DownloadUtilTest.GenerateFileName (don't think it's
> clang's fault)

I think all the downloads tests are broken, but we don't want to run
the bot until all the tests are green. That may mean disabling the
test.

> browser_tests: failed ExtensionApiTest.BookmarkManager

bug 5

[Nico Weber]

Awesome, we almost made it by end of 2010 :-)

I will roll the bots some time this week. Do you have a revision I
should roll to? Else I will just try with whatever's trunk when I do
it.

On Wed, Jan 5, 2011 at 10:53 AM, Evan Martin <ev...@chromium.org> wrote:
> To track work on all of this, I suggest filing bugs with an additional
> "Clang" label.

+1, I've used "clang" in the past for a few bugs as well.

>
> On Wed, Jan 5, 2011 at 5:47 AM, Hans Wennborg <hwen...@google.com> wrote:
>> I think we can now update http://code.google.com/p/chromium/wiki/Clang
>> to remove the notes about setting linux_use_tcmalloc=0 ?
>
> bug 1

Or just remove it, should take < 1 minute.

>
>> And I suppose we should also update the Linux bot to make all, and run
>> some more tests?
>
> bug 2

You can assign this to me.

[Hans Wennborg]

On Wed, Jan 5, 2011 at 6:53 PM, Evan Martin <ev...@chromium.org> wrote:
> To track work on all of this, I suggest filing bugs with an additional
> "Clang" label.

Will do.

>
> On Wed, Jan 5, 2011 at 5:47 AM, Hans Wennborg <hwen...@google.com> wrote:
>> I think we can now update http://code.google.com/p/chromium/wiki/Clang
>> to remove the notes about setting linux_use_tcmalloc=0 ?
>
> bug 1

Fixed.

>
>> And I suppose we should also update the Linux bot to make all, and run
>> some more tests?
>
> bug 2

crbug.com/68713

>
>> base_unittests: fails ProcessUtilTest.GetTerminationStatusCrash
>
> bug 3

crbug.com/68707

>
>> test_shell_tests: fails ICOImageDecoderTest.Decoding, and a few more
>
> bug 4

Hmm, these fail when I do a regular build too, so probably not Clang's fault.

>
>> unit_tests: fails DownloadUtilTest.GenerateFileName (don't think it's
>> clang's fault)
>
> I think all the downloads tests are broken, but we don't want to run
> the bot until all the tests are green.  That may mean disabling the
> test.
>
>> browser_tests: failed ExtensionApiTest.BookmarkManager
>
> bug 5

crbug.com/68711