Re: httex test fail info

[Patrice Dumas]

On Mon, Mar 19, 2012 at 08:15:41AM -0700, Karl Berry wrote:
> Hi Patrice,
> 
> For a few days now my nightly run of make check has been failing:

That's a very good idea to have a nightly run of make check!  Could it
be possible for you to run a 

  make long-check

too?  Once a week would certainly be enough.  One have to be in the 
tp/tests directory to run the long-check.

> The full output is below.  
>
> I wonder about using "F" (for fail) instead of "D".  I think it would be
> more intuitive.

There is a F followed by the file name which holds stderr result
when there is an exit code != 0.  I can also leave a F when a diff is
produced, if you like better.

> Also, I wonder about using "fail" or "ok" instead of "1" and "0" for the
> subtest results.

Ok, done.

> And including the names of the failed tests?  If it's easy.

I am not sure that would really help, as the failed tests are already
above.  It would clutter a lot in some cases.

> Anyway, the actual diff files are one line each:
> Only in out_parser/tex_httex: zztex_tex4ht_tex.ps
> and
> Only in out_parser/tex_complex_httex: zztex_complex_tex4ht_tex.ps
> 
> Offhand, I have no idea what's going on.  Any direction?

Your tex4ht version also created a .ps file when doing the html.  Also 
it is a bit weird because there is a zz prepended not present here. Did you
by chance change your tex4ht/texlive version (I assume you use texlive
;-)?  That looks a bit weird, maybe you could investigate what happens?
It is not completly straightforward to reproduce, because of the
indirect tex4ht call. Normally, to reproduce the tex4ht call, you
should, for example do

cd texinfo/tp/tests/formatting
cd out_parser/tex_complex_httex
httex tex_complex_tex4ht_tex.tex

Then you could try to understand why such a file appeared.  If you know
that it is normal, I can also handle the issue by not taking into
account the .ps when doing the diff.

-- 
Pat