Re: [Quilt-dev] Re: config-cc patch

[John Vandenberg]

On 8/16/05, Jean Delvare <address@hidden> wrote:
> Hi John,
> 
> [John Vandenberg]
> > I wonder what other small glitches are hidden away.  I dont think
> > diffstat is covered in the tests; any other tests that need to be
> > written?
>
> Since quilt 0.42, diffstat version is checked at ./configure time. I'd
> expect this to be sufficient, as there is only one diffstat, Thomas
> Dickey's one, right? So if we have the proper version is must be OK.
> 
> diffstat comes with its own test suite if you want to make sure it was
> properly compiled, so I see little benefit in adding a test for this to
> the quilt test suite.

I think it would be useful to have a test to make sure the diffstat
output appears where it should in the quilt patches, and that the awk
script in refresh.in works as expected.  Can someone familiar with the
code answer a few queries and take a look at the attached changes?

rev 1.45 allows diffstat output to begin with '#?' ; is this for
compatibility with earlier versions?

I see a comment on rev 1.33 "it did swap lines".  What line swapping
is done, and why?

If I grok it correctly, the block "BEGIN { split(diffstat, ds_arr,
/\n/) }" doesnt appear to be in use.

The pattern /^#? .* \| / could accidentally match a pipe symbol used
in the header; will the replacement work for the supported diffstat
versions?

Would anybody mind if quilt added a blank line between the header and
the output of diffstat if the last line of the header contained text?

Finally, does anyone object to a new simple test diffstat.test being
added as an optional test?  I have taken care to avoid including
complex histograms in the test case to avoid the test case breaking
when the diffstat algorithms are updated.
--
John

diffstat.patch
Description: Binary data