Contributing to the project
Dear GCC team, I would like to contribute to the project. I have a decent background in programming, in a number of languages on a number of platforms, but no specific experience in developing a compiler. I have a pretty good university education on processors, but no intimate knowledge of any particular instruction set. I know that approaching any big software project requires a humble attitude, so I'd really be ready to work on any task, even the most simple/boring one at first. Just to familiarise myself with the codebase. Would someone indicate some easy first project to work on? Please don't just redirect me to the list of open projects. Not sure whether that list is still valid, and I'd appreciate a more specific proposal anyway. Thanks in advance and regards, Enzo Vitale
Re: Contributing to the project
On Sun, Sep 23, 2012 at 02:26:30PM +0200, enzovit...@libero.it wrote:
> Dear GCC team,
>
> I would like to contribute to the project.
>
> I have a decent background in programming, in a number of languages on a
> number of platforms, but no specific experience in developing a compiler. I
> have a pretty good university education on processors, but no intimate
> knowledge of any particular instruction set.
You should first be able to compile GCC from its source code. Depending on your
software engineering skills and habits, this may be not so simple for you
(because GCC has a lot of dependencies). If using some Linux distribution
like Debian or Ubuntu, running "aptitude build-dep gcc-4.7" could help.
>
> I know that approaching any big software project requires a humble attitude,
> so I'd really be ready to work on any task, even the most simple/boring one
> at
> first. Just to familiarise myself with the codebase.
You might want to read the many good slides on http://www.cse.iitb.ac.in/grc/
in particular on
http://www.cse.iitb.ac.in/grc/gcc-workshop-12/index.php?page=slides
You could also play with MELT, in particular its probe (which is imperfect and
buggy)
on http://gcc-melt.org/ ; some of the slides also give you a picture of GCC, in
particular
http://gcc-melt.org/GCC-MELT-HiPEAC2012.pdf
[if interested, you could also participate to MELT which is also a branch of
GCC,
but there are many other things in GCC]
Of course, there are lot of material on GCC on the web (most of them outside of
http://gcc.gnu.org/ ...)
>
> Would someone indicate some easy first project to work on? Please don't just
> redirect me to the list of open projects. Not sure whether that list is still
> valid, and I'd appreciate a more specific proposal anyway.
There is an important first mandatory thing.
Follow the steps on http://gcc.gnu.org/contribute.html and in particular
take care of the legalese. Concretely, you need to have a paper signed between
your employer or university
and the FSF which transfer the copyright of all future contributions from your
part to the FSF.
Getting that document signed takes month[s], so please begin to act quickly.
Some few things. (After having initiated the legal process which takes time).
First, read some documentations; then read some source files of GCC (the issue
is to not
read a lot of code). GCC is too huge to be understoood by a single human. At
last, propose small patches.
Some parts of GCC could attract you more than others.
Don't forget to ask questions on mailing lists or IRC.
Have fun.
--
Basile STARYNKEVITCH http://starynkevitch.net/Basile/
email: basilestarynkevitchnet mobile: +33 6 8501 2359
8, rue de la Faiencerie, 92340 Bourg La Reine, France
*** opinions {are only mines, sont seulement les miennes} ***
Re: Contributing to the project
On Sun, 23 Sep 2012, enzovit...@libero.it wrote: I would like to contribute to the project. Great! Would someone indicate some easy first project to work on? Please don't just redirect me to the list of open projects. Not sure whether that list is still valid, and I'd appreciate a more specific proposal anyway. Since you asked for something specific, fixing this bug: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54583 seems likely to be a one-liner, but you need to find out where that line is needed (note that I can be wrong). Front-end (language specific code) warnings are a domain where bugzilla contains a number of issues that shouldn't be too hard to fix. Then you can try to investigate some middle-end bugs to get familiar with other parts of the code. One important element is to be motivated. If you start from something you find too boring, you may give up. If you get interested say in libstdc++, you may then be motivated to fix some C++ front-end bug that affects it, or some missed optimization where you noticed you had to write twisted C++ to get gcc to generate the best asm. -- Marc Glisse
Question about adding omp nodes in generic form ... (GCC OpenMP)
Hi, Originally, programmers add OpenMP's compiler directives into programs and then GCC will parse these OpenMP programs into the corresponding generic nodes. We want to automatically detect simple loop constructs from the generic forms. Then the loop constuct is replaced with the corresponding omp nodes. For some simple loop constructs,users will not add any OpenMP's directives and can use GCC to generate OpenMP programs. We implemented our idea on GCC 4.3.6. Our approach is to add a new function, modified_walk(), in the function c_genericize(). We have successfully detected loop and replaced it to the corresponding omp node. The generic tree and gimple tree from our approach and original OpenMP program are identical. When the loop body contains one statement, our implementation executes and the result is correct. However, when the loop body contains more than one statements, our implementation will get the segmentation fault (fail in the function ei_container()). How can we fix this problem? Is there any parameters which we forget to modify them? Thank you very much for your reply. Peng-Sheng
Re: bootstrap broken w disable-checking
On Sat, Sep 22, 2012 at 9:40 AM, Andreast Tobler wrote: > Hi all, > > while testing some patches fro Michael M, I noticed that disable-checking > seems broken on trunk. I saw this on x86_64-freebsd and powerpc64-freebsd. > > Is this issue already known? > If not, usual bug report with preprocessed source? Yes, please file a bugreport. Richard. > TIA; > Andreas > > Here the details: > > - > /export/devel/net/src/gcc/head/gcc/gcc/ipa-prop.c: In function 'bool > visit_ref_for_mod_analysis(gimple, tree, void*)': > /export/devel/net/src/gcc/head/gcc/gcc/ipa-prop.c:1842:1: error: mismatching > comparison operand types > visit_ref_for_mod_analysis (gimple stmt ATTRIBUTE_UNUSED, > ^ > unsigned int > bool > if (i.2_16 == 1) > > /export/devel/net/src/gcc/head/gcc/gcc/ipa-prop.c:1842:1: internal > compiler error: verify_gimple failed > -
Re: bootstrap broken w disable-checking
On 23.09.12 18:24, Richard Guenther wrote: On Sat, Sep 22, 2012 at 9:40 AM, Andreast Tobler wrote: Hi all, while testing some patches fro Michael M, I noticed that disable-checking seems broken on trunk. I saw this on x86_64-freebsd and powerpc64-freebsd. Is this issue already known? If not, usual bug report with preprocessed source? Yes, please file a bugreport. Thank you, filed as 54684. Andreas
Possible headsup on GCC testsuite PCH failures
Can someone double check this for me I am getting 99 PCH failures on the testsuite from GIT HEAD. Many thanks in advance, Aaron
gcc-4.8-20120923 is now available
Snapshot gcc-4.8-20120923 is now available on ftp://gcc.gnu.org/pub/gcc/snapshots/4.8-20120923/ and on various mirrors, see http://gcc.gnu.org/mirrors.html for details. This snapshot has been generated from the GCC 4.8 SVN branch with the following options: svn://gcc.gnu.org/svn/gcc/trunk revision 191654 You'll find: gcc-4.8-20120923.tar.bz2 Complete GCC MD5=c98966c830826f6ad3dd252607067282 SHA1=3cfbe877ffefb1ca3d3afbd6884c4a15d608d76d Diffs from 4.8-20120916 are available in the diffs/ subdirectory. When a particular snapshot is ready for public consumption the LATEST-4.8 link is updated and a message is sent to the gcc list. Please do not use a snapshot before it has been announced that way.
Re: Possible headsup on GCC testsuite PCH failures
Yes, this is a known problem: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54645 A patch was proposed: http://gcc.gnu.org/ml/gcc-patches/2012-09/msg01563.html Sorry to break the test. Dehao On Sun, Sep 23, 2012 at 4:56 PM, Aaron Gray wrote: > Can someone double check this for me I am getting 99 PCH failures on > the testsuite from GIT HEAD. > > Many thanks in advance, > > Aaron
Broken Math Resource Link on Your Site
Hi Administrator, I came across your website and wanted to notify you about a broken link on your page in case you weren't aware of it. The link on gcc.gnu.org/ml/gcc-help/1999-q3n/msg00261.html which links to http://www.cs.unb.ca/~alopez-o/math-faq/math-faq.html is no longer working. I've included a link to a useful page on Math Resources that you could replace the broken link with if you're interested in updating your website. Thanks for providing a great resource! Link: http://www.onlinebachelordegreeprograms.com/resources/bachelor-of-arts-in-math-math-resources/ Best, Alexandra Sawyer
