Re: src bfd/ChangeLog bfd/archive.c bfd/bfd-in2.h ...

[Thomas Schwinge <thomas at schwinge dot name>]

Hello!

On Fri, Oct 29, 2010 at 09:17:42AM -0700, H.J. Lu wrote:
> On Fri, Oct 29, 2010 at 9:10 AM, Joseph S. Myers
> <joseph@codesourcery.com> wrote:
> > On Fri, 29 Oct 2010, hjl@sourceware.org wrote:
> >> ? ? ? ld/testsuite/ld-elf: compress.exp compress1.s compress1a.d
> >> ? ? ? ? ? ? ? ? ? ? ? ? ? ?compress1b.d compress1c.d
> >
> > FAIL: ld-elf/compress1c
> >
> > appears for targets not supporting -shared; it should be marked
> > UNSUPPORTED for them, like various other -shared tests, instead.
> 
> I checked in this patch.
> 
>  2010-10-29  H.J. Lu  <hongjiu.lu@intel.com>
> 
> +	* ld-elf/compress1c.d: Only run for Linux targets.

> +++ b/ld/testsuite/ld-elf/compress1c.d
> @@ -2,6 +2,7 @@
>  #as: --compress-debug-sections
>  #ld: -shared
>  #readelf: -S --wide
> +#target: *-*-linux*

Hrm, in which way is this test specific to a Linux target?  In other
words, doing this is precisely what I'm suggesting *to avoid* in my patch
posted at <http://sourceware.org/ml/binutils/2010-10/msg00485.html>.
Can't we solve this and similar issues properly instead?  For example, by
adding infrastructure to whitelist / blacklist these tests based on
*real* specifics, which target: *-*-linux* is not, in my opinion.  Other
opinions?


Regards,
 Thomas

Attachment: signature.asc
Description: Digital signature