|This is the talk page for discussing improvements to the Clang article.|
|This article is of interest to the following WikiProjects:|
- Is it valid to say that Clang's goal is to replace GCC, period? It has different design goals. Comparison with pros and cons listed is even provided on CLangs site. — Preceding unsigned comment added by 188.8.131.52 (talk) 17:52, 7 April 2013 (UTC)
- I don't think it is valid to mark this as "Apple's fronted". yes, apple sponsors most of the development now but the code is open source and once version 1.0 is released I believe a wider community will be gathered.
- The article says the C support is complete. this is not true, there are still many bugs and missing features (va_arg missing on amd64 etc.) which results in clang not being able to compile many common C apps. I am also not sure why is clang "low" on importance, I believe it has a bright future, possibly replacing gcc as "the free C-family compiler" —Preceding unsigned comment added by 184.108.40.206 (talk) 21:45, 26 October 2008 (UTC)
And did Apple really want to? I can see how its useful to them, even more so if it's basing itself more on bsd/bsd licence rather than linux/gpl. I don't really get the point, they use gcc from non gpl programs in an accepted way, yet I read "but Apple is heavily funding Clang now because the GNU people put them in their sights wrt GCC. GNU played rough and got one over on NeXT (with GCC), but it's obvious that that lesson has been learned. Apple won't even allow any use of GPLv3 code, let alone integration into their system. This is what happens when you act anti-socially." so im confused. 220.127.116.11 (talk) 12:18, 21 July 2009 (UTC)
- Is it just me or is the English word 'clang' English pronunciation: /ˈkleɪŋ/, and not English pronunciation: /ˈklæŋ/? KMeyer (talk) 02:36, 17 October 2009 (UTC)
- The article seems correct for me because I can imagine a fake word clayng English pronunciation: /ˈkleɪŋ/ that I wouldn't pronounce like clang. Googling found some brief mentions of [æŋ] becoming [eɪŋ] in parts of North America, which could be a possibility. —Mrwojo (talk) 03:43, 17 October 2009 (UTC)
- It seems strange that the Background section doesn't mention gcc's change of license (GPL2->GPL3) as a main reason for Apple dropping gcc. —Preceding unsigned comment added by 18.104.22.168 (talk) 20:16, 8 May 2011 (UTC)
- GPL3 also the reason for the BSDs moving to clang, for example FreeBSD has GPL3 versions of gcc in its ports system, but its base system is built with a GPL2 version from 2007. — Preceding unsigned comment added by 22.214.171.124 (talk) 18:31, 29 May 2011 (UTC)
Google uses Clang
A Google programmer has posted an interesting article about how Google uses and improves Clang for their C++ code. See either the LLVM blog or Google's Engineering Tools blog. They use Clang "to provide accurate and helpful warnings and diagnostics to engineers". They add warnings to Clang (ie., enhance it to detect "dangerous and potentially buggy constructs") and customize some of those warnings. They use these enhancements to systematically scan their entire code base:
- When other teams at Google respond to production bugs, our team will often begin working to enable any Clang diagnostics that might have caught the bug. Within one week of production issues, we can sweep the entire code base using these diagnostics to fix any latent bugs.
Should we mention Google's use of Clang in the article? (Any mention would need to be brief.) CWC 10:56, 3 June 2011 (UTC)
diffing the IF?
The Overview sections contains: "Changes to the compiler can be checked by diffing the IF.", yet IF is nowhere cited what it stands for. Can anybody knowledgable explain what IF is? Thanks, 126.96.36.199 (talk) 21:10, 23 July 2011 (UTC)
...runtime performance of clang/LLVM output is sometimes worse than GCC's. GCC's compilation speeds are really slow, but runtime performance of binaries is generally (9/10 itw cases) better by 10+%. Phoronix this tests can back me up. "sometimes" suggests that it's like that in <50% cases. I'd suggest replacing it with "mostly" or "generally" (with reasonable defaults such as "-O2" or even "-mntune=native" flags). — Preceding unsigned comment added by 188.8.131.52 (talk) 18:51, 13 April 2012 (UTC)
- Can you post a link to such survey? I then would update the article. 1exec1 (talk) 20:27, 23 April 2012 (UTC)
In article this is mentioned several times, but as of now (Aug '12) there is no incremental compilation option (standalone server instance, nor IDE support, not counting proprietary X code). I guess article should reflect this more, to me it sounds like failed design goal (my POV, not developer's ofc). — Preceding unsigned comment added by 184.108.40.206 (talk) 13:14, 12 August 2012 (UTC)