Re: [PATCH] Merge dg-extract-results.{sh,py} from GCC upstream

[Pedro Alves <palves@redhat.com>]

On 02/12/2015 08:32 PM, Sergio Durigan Junior wrote:
> On Thursday, February 12 2015, Pedro Alves wrote:
> 
>>> gdb/testsuite/ChangeLog:
>>> 2014-12-14  Sergio Durigan Junior  <sergiodj@redhat.com>
>>>
>>> 	Merge dg-extract-results.{sh,py} from GCC upstream (r210243,
>>> 	r210637, r210913, r211666, r215400, r215817).
>>>
>>> 	2014-05-08  Richard Sandiford  <rdsandiford@googlemail.com>
>>> 		* dg-extract-results.py: New file.
>>
>> Unfortunately, the Python version of this script (dg-extract-results.py)
>> doesn't produce stable results.  The sorting changes between runs for
>> some reason.  That may be tolerable for GCC, but for GDB, it often
>> renders test diffing unworkable.
> 
> Sorry for not catching this before.
> 
> Our BuildBot does not suffer from this because it doesn't rely on the
> order of the results in the .sum file...
> 
>> I think we should remove dg-extract-results.py, and re-sync
>> dg-extract-results.sh from the gcc repo to pick up your fix for
>> the grep/binary issue that went in meanwhile.
>>
>> WDYT?
> 
> I'm in favor.  But just to be safe (sorry for repeating): we need to
> pick up my fix from GCC.  Otherwise the BuildBot wil be badly affected.

Wait, we may have gotten things backwards.

Aren't we using the .sh version still?

testsuite/Makefile.in:
check-parallel:
	-rm -rf cache outputs temp
	$(MAKE) -k do-check-parallel; \
	$(SHELL) $(srcdir)/dg-extract-results.sh \
	  `find outputs -name gdb.sum -print` > gdb.sum; \
	$(SHELL) $(srcdir)/dg-extract-results.sh -L \
	  `find outputs -name gdb.log -print` > gdb.log
	@sed -n '/=== gdb Summary ===/,$$ p' gdb.sum

Now I'm confused.  Is the issue then that patches to the .sh version
that we pulled in with the last merge introduced the bad sorting?

/me confused.

Thanks,
Pedro Alves