facebook warp

閲覧: 59 回
最初の未読メッセージにスキップ

Thiago Farina

未読、
2015/02/25 12:56:142015/02/25
To: Chromium-dev
Has anyone looked into using warp preprocessor instead of gcc's preprocessor?

They seem to have achieved good results by using it [1].

Jeremy Roman

未読、
2015/02/25 13:51:432015/02/25
To: Thiago Farina、Chromium-dev
I haven't looked deeply into it, but at the time it was announced, I heard that Clang still outperformed it.
https://news.ycombinator.com/item?id=7489532 (the author of this post is Google's Daniel Berlin, I believe).

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev

Nico Weber

未読、
2015/02/25 14:01:092015/02/25
To: Jeremy Roman、Thiago Farina、Chromium-dev
warp is meant as a replacement for gcc -E, the preprocessor-only mode of the compiler. When using distcc, one needs to quickly figure out which files to upload to the build servers ( http://google-opensource.blogspot.com/2008/08/distccs-pump-mode-new-design-for.html ), and one way to do this is to preprocess the files and see which #includes get processed. But this doesn't require a "full" preprocessor, and I think this is what warp does. (goma also has its own preprocessor for performance reasons, and it's faster than a "real" preprocessor because it does less work.)

So if you're using distcc, this might be useful for you. If you're building locally, this isn't for you.

Nico

Reid Kleckner

未読、
2015/02/25 14:09:432015/02/25
To: Jeremy Roman、Thiago Farina、Chromium-dev
The main concern for Chromium is that Clang's diagnostics are different after macro expansion. We use this to suppress low-value or false positive warnings in some corner cases. Some users argue that this is a mistake in Clang, but it doesn't change the fact that we can't use warp in Chromium today without triaging a load of warnings.

Clang also provides the -frewrite-includes flag, which prevents macro expansion and only expands includes. I think goma uses it. Warp would need to implement this functionality before we could consider it for Chromium.

On Wed, Feb 25, 2015 at 10:50 AM, Jeremy Roman <jbr...@chromium.org> wrote:
To unsubscribe from this group and stop receiving emails from it, send an email to chromium-dev...@chromium.org.

全員に返信
投稿者に返信
転送
新着メール 0 件