Re: bootstrap differences

[Eric Blake]

According to Jim Meyering on 2/23/2010 2:10 AM:
> Eric Blake wrote:
>> Should any of these bootstrap differences be synchronized?  Both coreutils
>> and gnulib had some improvements over the other:
>> @@ -594,7 +594,6 @@
>>        (
>>          echo "$c1-*- buffer-read-only: t -*- vi: set ro:$c2" &&
>>          echo "${c1}DO NOT EDIT! GENERATED AUTOMATICALLY!$c2" &&
>> -        echo '#line 1' &&
> 
> The use of "#line 1" in coreutils is not worth keeping.
> At first I thought it would be, but in practice it just causes trouble,
> at least for me, via the off-by-3 line numbers in failed assertions.

It all depends on whether you are looking in pristine gnulib or in the
coreutils modified copy.  One of the two versions will have line numbering
off, so it boils down to a question of which source people are more likely
to inspect during a bug report.  My guess is that most reports will be
from the package perspective and not to gnulib, particularly since the
line numbers frozen at the time of the package tarball creation are more
important than the current line numbers in upstream gnulib at the time of
the bug report.  Therefore, I'm inclined to agree with your assessment
that the #line 1 is not as helpful.  So I'm leaving that out of gnulib.

> 
>> +# Local Variables:
>> +# indent-tabs-mode: nil
>> +# End:
> 
> These lines (in gnulib's copy) are no longer needed,
> now that we have an indent-with-spaces policy.

Then I'm pushing this to drop them from gnulib, along with some comment
cleanups.

-- 
Eric Blake   address@hidden    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

gnulib.patch344
Description: Text document

signature.asc
Description: OpenPGP digital signature