Review Request: __attribute__((...)) parsing fixed with MinGW headers

[Milian Wolff]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/104805/#review13311
-----------------------------------------------------------


hm considering that __attribute__ is not part of the standard anyways, removing it from our parser and workarounding it with such an empty macro sounds fine to me.

please add a unit test in cpp/tests/test_cppcodecompletion.cpp (ignore the file name...). there you have e.g. testPreprocessor() which shows you how to test this, e.g. by assuring

a)
#ifdef __attribute__
#define FOO 1
#else
#define FOO 0
#endif
FOO
=> preprocessed == 1
b) 
#undef __attribute__
__attribute__(1)
=> preprocessed == "" (empty)

- Milian Wolff


On May 1, 2012, 11:36 a.m., Ivan Shapovalov wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/104805/
> -----------------------------------------------------------
> 
> (Updated May 1, 2012, 11:36 a.m.)
> 
> 
> Review request for KDevelop.
> 
> 
> Description
> -------
> 
> MinGW system headers tend to undefine __attribute__ (which is internally defined to no-op in the KDevelop preprocessor since C++ parser cannot handle it correctly in all cases).
> 
> Patch just adds "fixed" property to the definition.
> Also, removed from parser some dead code related to __attribute__ parsing, since it handles only one case from possibly infinite number of these.
> 
> 
> Diffs
> -----
> 
>   languages/cpp/parser/lexer.cpp d097710 
>   languages/cpp/parser/parser.cpp 172b6b6 
>   languages/cpp/parser/tokens.h 77ab417 
>   languages/cpp/parser/tokens.cpp 7df4bb9 
>   languages/cpp/setuphelpers.cpp 59bb850 
> 
> Diff: http://git.reviewboard.kde.org/r/104805/diff/
> 
> 
> Testing
> -------
> 
> Used it with MinGW and normal system headers, no regressions detected. Unit-tests are passed.
> 
> BTW: I couldn't find out how to use the standard preprocessing environment (which has been modified) in preprocessor unit-tests, so no unit-tests added.
> 
> 
> Thanks,
> 
> Ivan Shapovalov
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20120503/244f611a/attachment.html>