posix-c99-c11 coverage

posix-c99-c11 coverage

[Philippe Lavoie]

[-- Attachment #1: Type: text/plain, Size: 354 bytes --]

Hi,

We are porting musl to our dsp architecture.

Since there are a quite a few functions we do not implement, we would like to generate a compatibility table like the ones listed at http://wiki.musl-libc.org/wiki/Compatibility for posix, c99 and c11.

Is there a script or test suite (libc-test?) to generate the tables?

Thanks,
-Philippe


[-- Attachment #2: Type: text/html, Size: 917 bytes --]

Re: posix-c99-c11 coverage

[Rich Felker]

On Thu, Dec 18, 2014 at 09:30:42PM +0000, Philippe Lavoie wrote:
> Hi,
> 
> We are porting musl to our dsp architecture.
> 
> Since there are a quite a few functions we do not implement, we
> would like to generate a compatibility table like the ones listed at
> http://wiki.musl-libc.org/wiki/Compatibility for posix, c99 and c11.
> 
> Is there a script or test suite (libc-test?) to generate the tables?

I think they're generated from the code in this repo:

http://nsz.repo.hu/git/?p=musl-tables;a=tree

Rich

Re: posix-c99-c11 coverage

[Szabolcs Nagy]

* Rich Felker <dalias@libc.org> [2014-12-18 16:44:48 -0500]:
> On Thu, Dec 18, 2014 at 09:30:42PM +0000, Philippe Lavoie wrote:
> > 
> > Is there a script or test suite (libc-test?) to generate the tables?
> 
> I think they're generated from the code in this repo:
> 
> http://nsz.repo.hu/git/?p=musl-tables;a=tree
> 

these are messy shell+awk scripts
(never intended for others to look at it)

basic idea:

- data dir has c99, c11 and posix2008 api list
(tab separated values, c api is from annex b of the iso pdf,
posix api is from the downloadable html)

- defined symbols in libc.a are listed with nm

- declared symbols in headers are listed with ctags
(this is not perfect, will parse all ifdef branches etc
and i try to check the prototypes too which is hard)

- musl syms are compared to the api lists

furthermore libc-test has an api test which does check
function prototypes properly for all posix functions
(but there is no nice report about it)

and i have some scripts to test posix namespace violations
using ctags data and the posix namespace rules which i
collected here:
http://port70.net/~nsz/c/posix/

hope this helps