Problem with building PDFs

[Joel Sherrill]

On 01/23/2013 01:20 PM, Freddie Chopin wrote:
> W dniu 2013-01-23 17:47, Joel Sherrill pisze:
>> _Exit.def was empty for me when "make pdf" failed. I removed it
>> by hand and did another make and it succeeded. On another run,
>> there was another .def file which was empty.
> But if I remove this file (which is generated during "make pdf") and do
> another "make pdf" it is regenerated and the error is the same...
>
>> When I did a "make info pdf", it seemed to work more reliably.
> Indeed "make info pdf" "works", but the end result is the same as
> previously - the .pdf is created, yet the process exits with error /;
>
> First run:
>
>> Output written on libc.pdf (326 pages, 695631 bytes).
>> Transcript written on libc.log.
>> /usr/bin/texi2dvi: pdfetex exited with bad status, quitting.
>> make[2]: *** [libc.pdf] Error 1
> Second run:
>
>> Output written on libm.pdf (98 pages, 356790 bytes).
>> Transcript written on libm.log.
>> /usr/bin/texi2dvi: pdfetex exited with bad status, quitting.
>> make[2]: *** [libm.pdf] Error 1
> Third run is OK, as the PDFs are there.
>
> But your message made me do a little "search" and indeed there are some
> empty (0 bytes) files called stmp-def in many folders + some "normal"
> file that are empty:
>> [freddie@arch newlib]$ find -size 0
>> ./libm/math/stmp-def
>> ./libm/complex/stmp-def
>> ./libm/stmp-targetdep
>> ./libm/common/stmp-def
>> ./libc/stmp-iconvset
>> ./libc/time/stmp-def
>> ./libc/string/stmp-def
>> ./libc/stmp-sigset
>> ./libc/stmp-stdio64
>> ./libc/stdio/getwc.def
>> ./libc/stdio/putwc.def
>> ./libc/stdio/stmp-def
>> ./libc/signal/stmp-def
>> ./libc/stdlib/mallocr.def
>> ./libc/stdlib/stmp-def
>> ./libc/misc/stmp-def
>> ./libc/ctype/stmp-def
>> ./libc/stmp-posix
>> ./libc/stmp-targetdep
>> ./libc/search/stmp-def
>> ./libc/reent/stmp-def
>> ./libc/stmp-extra
>> ./libc/locale/stmp-def
>> ./stmp-targ-include
> So three .def files that are empty. Could that be the reason? These 3
> names cannot be found in the log of "make pdf", so I guess not...
No. The *.def files that are empty are the problem. They are
not being generated properly.
> I guess the real reason is somewhere here:
> I've got a failure for these 5 files:
> libc:
> sscanf.def:275
> swscanf.def:261
> strftime.def:212
> libm:
> isgreater.def:55
> s_matherr.def:81
> Each failure is in exactly the same place:
Those files include references to "nodes" which are supposed
to be defined in the empty "*.def" files. So the linkage is broken
before some nodes are empty.

The empty *.def files are triggering the failures. Why the build
is generating them empty is the question.
>> @end table
>>
>>
>> @*<- here's the failing line
> I bet such code is present only in these 5 files and nowhere else. In
> each case the error is the same:
>
>> You can't use `@unskip' in vertical mode.
>> @*->@unskip
>>              @hfil @break @hbox {}@ignorespaces
>> l.81 @*
> (the line number changes of course [; )
>
> If I remove the troubling @* the error is gone...
>
> Is there anything wrong with this LaTeX construct or maybe my LaTeX is
> misbehaving?
>
> 4\/3!!


-- 
Joel Sherrill, Ph.D.             Director of Research&  Development
joel.sherrill@OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35806
Support Available               (256) 722-9985