[texdoc] [tex-live] texdoc + AR9 on win32

[Reinhard Kotucha]

Philip TAYLOR (Ret'd) writes:

 > 	os.execute('Rundll32 Shell32,ShellExec_RunDLL c:\\tmp\\file.pdf')
 > 
 > Thus everything seems to point at kpse.set_program_name('texlua')
 > as being the culprit; who can look inside kpse.set_program_name ?

Hi Phil,
yes, Manuel mentioned this on one of his mails already.  From all I've
seen so far, I deduce that we can be sure now that AR does *not*
crash when it is processing the argument vector.  No, the crash
happens later and we should have recognized this *much* erarlier.  

I always had Fengnan's message from Aug 31 in mind but I never had
been able to draw any conclusions from it.

 > This time I'm going to report something more I find today about the
 > last email I sent to you. I happened to find that if I keep a Acrobat
 > window open( it doesn't matter whether it is active), commands like
 >     texdoc minitoc
 > works fine. However if there is no acrobat.exe loaded in memory, the
 > command will not work.

This is very useful hint.

Arguments are passed to a program before it can do anything useful.
This means that it gets its argument vector before it knows whether it
is already running.  It always gets the same argument list.  I suppose
that there is nothing wrong with os.spawn() in this respect.

But maybe there is something we neglected.  We concentrated on the
argument vector but maybe the problem is caused by the environment.

Phil, if you say 'start foo.pdf' then start associates .pdf with
acrod32.exe.  Suppose that if you replace foo.pdf by foo.bat, cmd.exe
will be invoked.  Right?

Then you can explore the environment easily if the batch file contains 

   set > c:\temp\env.txt

It is extremely helpful to know how the environment looks like.

And it is extremely helpful to find out whether kpse.set_program_name
is the culprit.  Maybe it changes the environment.  

I'm quite convinced that there is something wrong with luatex.

Regards,
  Reinhard

-- 
----------------------------------------------------------------------------
Reinhard Kotucha			              Phone: +49-511-3373112
Marschnerstr. 25
D-30167 Hannover	                      mailto:reinhard.kotucha at web.de
----------------------------------------------------------------------------
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.
----------------------------------------------------------------------------