Move make_clang_dir handling later so --print-revision works

Move make_clang_dir handling later so --print-revision works

R=thakis@chromium.org
BUG=

Committed: https://chromium.googlesource.com/chromium/src/+/1d3d4c44705cba77dc133b47ba2b7e119a0ab60e

[Reid Kleckner, 2015-07-10 17:07:02]

Hm, next time I should pass --send-mail.

[Nico, 2015-07-10 17:08:37]

lgtm

but I wonder if we can just delete that block altogether. now that we have
prebuilts, there's almost no cost to just updating the clang in
third_party/llvm-build too, even if you end up using a different clang

[Reid Kleckner, 2015-07-10 17:12:32]

On 2015/07/10 17:08:37, Nico wrote:
> lgtm
> 
> but I wonder if we can just delete that block altogether. now that we have
> prebuilts, there's almost no cost to just updating the clang in
> third_party/llvm-build too, even if you end up using a different clang

Maybe.

I noticed this change doesn't actually do the right thing for new warning flags.
--print-revision thinks it's using the prebuilt clang, which doesn't know about
-Wshift-negative-value, and I get build errors later.

[Nico, 2015-07-10 17:19:11]

On Fri, Jul 10, 2015 at 10:12 AM, <rnk@chromium.org> wrote:

> On 2015/07/10 17:08:37, Nico wrote:
>
>> lgtm
>>
>
>  but I wonder if we can just delete that block altogether. now that we have
>> prebuilts, there's almost no cost to just updating the clang in
>> third_party/llvm-build too, even if you end up using a different clang
>>
>
> Maybe.
>
> I noticed this change doesn't actually do the right thing for new warning
> flags.
> --print-revision thinks it's using the prebuilt clang, which doesn't know
> about
> -Wshift-negative-value, and I get build errors later.
>

Yes, I figured you'd tweak build flags locally too. I guess this script
could try to get the revision from the clang binary in make_clang_dir, but
it seems like quite a bit of complexity for the number of users of this
flag.

(This particular problem will go away after the next clang/win roll)

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Reid Kleckner, 2015-07-10 23:23:17]

Committed patchset #1 (id:1) manually as
1d3d4c44705cba77dc133b47ba2b7e119a0ab60e (presubmit successful).