From bug-octave-request at che dot utexas dot edu  Thu Oct 20 11:57:25 1994
Subject: some problems
From: mguttman at mri dot jhu dot edu
To: bug-octave at che dot utexas dot edu
Cc: eatalar
Date: Thu, 20 Oct 94 12:54:25 -0400


Hello:

1) A comment:
    Thank you (and others) for writing this program!
    Other options are very expensive, and octave seems
    to be quite nice and capable.  I would like to try
    linking it with SGIs parallelized versions of the BLAS,
    EISPACK and LINPACK libraries.  These libraries are
    thrown in when you buy FORTRAN.

2) A bug:
    I compiled octave 1.0 on IRIX 5.2 with gcc 2.5.8.
    As far as I can tell, everything works OK except for "quit".
    This causes a crash and the dialog is below.
    -----------------------------------------------------------
    Octave, version 1.0.  Copyright (C) 1992, 1993, 1994 John W. Eaton.
    This is free software with ABSOLUTELY NO WARRANTY.
    For details, type `warranty'.

    octave:1> quit
    error: caught SIGSEGV -- stopping myself...
    DLList Error: null Pix
    error: caught SIGIOT -- stopping myself...
    DLList Error: null Pix
    IOT trap (core dumped)
    ------------------------------------------------------------

3) A question:
    Octave seems not to have any minimization functions
    (a la Matlab's fmin and fmins).  Am I looking in the
    wrong place?

4) A bug related to the question:
    We were pleasantly surprised to see that octave reads
    Matlab '.m' files with out alteration.  So, I thought,
    "Why not copy over Matlab's fmin.m and related files
    and give it a try.  This is what happened:
    ----------------------------------------------------------------
    octave:5> fmin('cos',3,4)
    tc-index.cc:363: failed assertion `(nr == 1 || nc == 1) && !
user_pref.do_fortran_indexing'
    error: caught SIGIOT -- stopping myself...

    error: caught SIGSEGV -- stopping myself...
    DLList Error: null Pix
    IOT trap (core dumped)
    ----------------------------------------------------------------
    I can understand the script not working, but it shouldn't crash.

5) A mistake in scripts/Makefile?
   It should perhaps have this diff:
----------------------------------------
45c45
<         rm -f $(libdir)/$$f ; \
---
>         rm -f $(libsubdir)/$$f ; \
----------------------------------------

6) A bug?
   I specified another directory to use for the info files and
   it doesn't find them.  I'm doing this because I have a
   multi-architecture lab and thought I would be clever about sharing
   info files and script files.

I gave up and got the IRIX 4.x binary.

Mike Guttman