Bug 194166 - [code model + code completion + discovery] Support Clang+LLVM
[code model + code completion + discovery] Support Clang+LLVM
Status: NEW
Product: cnd
Classification: Unclassified
Component: Code Model
PC Linux
: P3 with 13 votes (vote)
Assigned To: issues@cnd
Depends on:
  Show dependency treegraph
Reported: 2011-01-09 21:35 UTC by rmartins
Modified: 2012-12-03 21:30 UTC (History)
2 users (show)

See Also:


Note You need to log in before you can comment on or make changes to this bug.
Description rmartins 2011-01-09 21:35:31 UTC
I think the support brought my clang/llvm tool-chain would greatly improve the performance + accuracy of CND's parser, specially on templates. Given that it was built to be tightly binded to IDE's, supporting it shouldn't be a total nightmare;)


Comment 1 danleepw 2012-06-28 07:06:16 UTC

any update about this enhancement?

Comment 2 Vladimir Voskresensky 2012-06-28 07:23:27 UTC
(In reply to comment #1)
> Hi,
> any update about this enhancement?
> Regards,
> Dan
Hi Dan, 
which part of this issue you are interested in?
Do you have concerns about parsing speed? Completion accuracy?

Or may be you are interested in clang compiler support?

Comment 3 danleepw 2012-06-28 07:26:32 UTC
Hi Vladimir,

Nice to talk with you again. To your questions, i am actually interested in the parsing speed and accuracy, but not the compilation support.

Comment 4 Vladimir Voskresensky 2012-06-28 19:39:27 UTC
(In reply to comment #3)
> Hi Vladimir,
> Nice to talk with you again. 
Me too :-)
> To your questions, i am actually interested in the
> parsing speed and accuracy, but not the compilation support.
What's the size of your project? How long does it take to parse in 7.2?
Can you clarify accuracy problems? Do you mean unresolved identifiers or bad completion choices? May be you face the instability when some files after editing become polluted with unexpected unresolved identifiers.
Comment 5 danleepw 2012-06-29 04:37:04 UTC
Hi Vladimir,

My project size is around 950K (just includes source code). You are right that I am facing the instability that unresolved identifiers are often appearing, even I just open the source code which hasn't been edited recently. Also, the code completion is somehow very slow when the project gets large.

I have already put the cache directory into a ram disk. It does give some improvement but still not fast enough.

Comment 6 rmartins 2012-07-05 14:12:51 UTC
Hi Vladimir and Dan,
I use netbeans in a relative small project, 50k SOC within 1000 files, 
and despite its major improvements (thanks Vladimir), it is still sluggish. 
IMHO netbeans has to provide an alternative backend for the parsing, either 
using clang/llvm tool-chain or by having a native (portable) C/C++ parser.
Perhaps the later would provide a more flexible approach.

Comment 7 yuriy_lalym 2012-08-26 12:14:50 UTC
(In reply to comment #3)
QtCreator vs QtCreator-wip/clang. I compare the time spent on analysis - the first is much faster. But accuracy of the analysis in the second case the very good. If you only want to receive high speed of parsing that clang most likely won't help.

И огромное спасибо за NetBeans 7.2
Comment 8 chila 2012-12-03 21:30:41 UTC
Wouldn't it be faster to use a compiler to analyze the code rather than java code?. Code completion in Netbeans is slow, too memory consuming and sometimes unaccurate. Also, there'll be no need to duplicate analysis code. The analysis in clang is very good, better than gcc.

By use of this website, you agree to the NetBeans Policies and Terms of Use. © 2014, Oracle Corporation and/or its affiliates. Sponsored by Oracle logo