Seems like a smart idea and solid patch. I have a couple suggestions you
can do with as you please:

On Mon, Jun 03, 2013 at 04:33:35PM -0400, Peter Harris wrote:
> If an installed xtest is run, the user does not normally have write
> access to the test directory. Writing intermediate files and results
> directly to the results directory avoids a number of spurious FAILs and
> UNRESOLVEDs.
...
> +const char *
> +outfile(const char *fn)
> +{
> +     char *out = (char *)fn;

This assignment to out is dead; I'd delete the initializer.

> +     const char *path;
> +     char *resfile = getenv("TET_RESFILE");
> +     if (!resfile)
> +             return fn;
> +
> +     resfile = strdup(resfile);
> +     if (!resfile)
> +             return fn;
> +
> +     path = dirname(resfile);
> +     if (!path)
> +             goto done;
> +
> +     out = malloc(strlen(path) + strlen(fn) + 2);
> +     if (!out) {
> +             out = (char *)fn;
> +             goto done;
> +     }
> +     regid(NULL, (union regtypes *)&out, REG_MALLOC);

I don't know the XTS codebase (and don't want to, but...): Does regid
register a resource to be freed at the end of the current test?

- If so, this patch seems clearly correct and a good idea. A comment at
  the declaration of outfile() about the lifetime of the result would be
  nice though.

- If not, meh, it won't leak so much it's a serious problem.

Either way:
Reviewed-by: Jamey Sharp <[email protected]>

> +     sprintf(out, "%s/%s", path, fn);
> +
> +done:
> +     free(resfile);
> +     return out;
> +}

Jamey

Attachment: signature.asc
Description: Digital signature

_______________________________________________
[email protected]: X.Org development
Archives: http://lists.x.org/archives/xorg-devel
Info: http://lists.x.org/mailman/listinfo/xorg-devel

Reply via email to