From maintainers-request at octave dot org Sun Jan 30 15:41:26 2005
Subject: Re: Reprise: external pager quitting kills standard error
From: Daniel J Sebald <daniel dot sebald at ieee dot org>
To: maintainers at octave dot org
Cc: maintainers at octave dot org
Date: Sun, 30 Jan 2005 15:47:22 -0600

I searched around a bit for the reason the error messages are lost.  
Although I see a "nonconformant matrices" and similar in the /src 
subdirectory, it appears that it is the "gripes_#" that is handled in 
the /liboctave subdirectory because that is where "nonconformant 
arguments" appears and that is what is displayed by Octave.

I really haven't found a reason why the error reporting would be lost.  
One peculiar thing I see though is that only in cmd-edit.cc is 
current_liboctave_error_handler called in this fashion:

      current_liboctave_error_handler
    ("unable to create command history object!");

In all other (so many) instances in the /liboctave subdirectory it is 
called as:

  (*current_liboctave_error_handler)
    ("%s: nonconformant arguments (op1 len: %d, op2 len: %d)",

My mind isn't working well enough right now to think what the difference 
is between calling by reference and calling by pointer,  In any case, 
perhaps changing cmd-edit.cc to (*current_liboctave_error_handler") form 
would be good for consistency.

Dan

PS:  After a complete search of the tree, I found the definition of 
these things in the ominous sounding "libcruft/misc" subdirectory:

typedef void (*liboctave_error_handler) (const char *, ...);

/* Would be nice to make these pointers private, but we want to share
   them among all the liboctave classes. */
extern liboctave_error_handler current_liboctave_error_handler;





Daniel J Sebald wrote:

> (please cc: any replies, I'm not on list)
>
>
> Has someone found a solution to the pager problem in this archived
> report from March?
>
>
> http://www.octave.org/octave-lists/archive/bug-octave.2004/msg00200.html
>
> http://www.octave.org/octave-lists/archive/bug-octave.2004/msg00201.html
>
>
> I'm running Fedora Core 3, which is probably a derivative of RH 9.0, so
> perhaps the problem is still propagating through RedHat's distribution.
>
> Note that this doesn't happen with a smaller amount of data in "less",
> only when there are larger amounts of data in "less".  It's as though
> some pipe buffer may be overflowing.  Here's example output where
> ones(25) displays fine, but ones(100) goes awry.
>
> octave:1> bogus
> error: `bogus' undefined near line 1 column 1
> octave:1> junk = ones(25)
> octave:2> bogus
> error: `bogus' undefined near line 2 column 1
> octave:2> junk = ones(100)
> warning: connection to external pager (pid = 18461) lost --
> warning: attempting to finish pending computations...
> octave:3> bogus
> octave:3>
>
> Thanks,
>
> Dan
>
>
>


-- 

Dan Sebald
email: daniel DOT sebald AT ieee DOT org
URL: http://acer-access DOT com/~dsebald AT acer-access DOT com/