On 12/17/2010 01:39 PM, Caspar Zhang wrote:
> Hi list, I open this thread to discuss the plan and implementation of
> improving crackerjack code quality, anyone who is interested in it is
> welcomed to join the discussion.
>
> What I've met for now are as follows:
>
> 1) Handling non-existed syscalls. I wonder if anyone got a chance to
> look into LTP's codes, maybe we can backport some codes from it to
> handle this issue. Anyway, I think it is necessary to handle non-existed
> syscalls, else the compilation will not pass.
>
> 2) feature test macros. I don't know much about that, but I guess
> _GNU_SOURCE macro is necessary, while others like: SUSv2 (UNIX 98) which
> needs _XOPEN_SOURCE defined and value > 500, is it necessary? does
> crackerjack supposed to be run on Unix systems?
fixed one in r1004, will fix them if I find more similar defects.
>
> 3) hard-coded syscalls. take ssetmask testcase, in Line 25:
> syscall(69, SIGALRM);
>
> Different arch has different number for ssetmask() syscall, we shouldn't
> use hard-coded number.
>
> 4) Windows ^M character and non-English comments: see delete_module03
> testcase.
fixed in r999, will fix them if I find more similar defects.
>
> That's all so far, will add more items if I find.
Thanks,
Caspar
--
Quality Assurance Associate (Kernel) in
Red Hat Software (Beijing) Co., R&D Branch
TEL: +86-10-62608150
WEB: http://www.redhat.com/
|
