Subject: need volunteer for scilab pkg test
To: None <email@example.com>
From: None <firstname.lastname@example.org>
Date: 08/18/1998 18:01:20
I am pleased to announce that I think I finally have a scilab-2.4 package
done. Scilab is very much like matlab. Its an interactive programming
language with scientific math emphasis. The program contains extensive
math libraries and high quality X11 graphics. See
http://arikara.inria.fr/www-rocq.inria.fr/scilab/scilab.html for more info
I have tested the package on NetBSD-1.3.2/pmax and
it appears to be fully operational. There are 2 bugs under
NetBSD-1.3/mac68k however. As my mac is in a highly critical path right
now, I'm dragging my feet on upgrading to 1.3.2 to see if the problem goes
away. If anyone has a 1.3.2 or -current mac and about 100MB of free disk
space, I'd appreciate it if they could try it out and let me know the
results. Anyone who wants to try it out needs to follow steps 1-5 below.
The other item which I could use some testers on is the following. Scilab
includes the ability to dynamically link in C and FORTRAN routines. This
feature works on all the NetBSD systems I've been able to test it on
(NetBSD-1.3/mac68k and NetBSD-1.3.2/pmax and i386), but I don't know if
perhaps there will be some problems on other platforms (alpha maybe?).
Anyone wanting to try this part needs to do steps 1-4 below.
0) you will need f2c and xless. There are packages for these.
1) Download the pkgsrc
# cd /tmp
# ftp http://www-mtl.mit.edu/~mcmahill/netbsd/scilab-2.4-pkgsrc.tar.gz
2) extract pkgsrc
# cd /usr/pkgsrc/math
# gunzip < /tmp/scilab-2.4-pkgsrc.tar.gz | tar xvf -
3) build the package
# cd /usr/pkgsrc/math/scilab
# limit datasize 40000
4) run the built in tests
# cd /usr/pkgsrc/math/scilab/work/scilab-2.4/tests
# make tests 2>&1 | tee make_test.log
examine the file
and verify that everything passed. If something failed, please send me
the message. My 1.3 mac fails some, haven't figured out yet why, but
they don't appear to be serious (see the README in that directory).
The 1.3.2 pmax passes all with flying colors as does 1.3.2 i386.
5) run my extra tests which failed on my NetBSD-1.3/mac68k
This will bring up the main scilab window. You should see something
S c i l a b
Scilab-2.4 (July 12, 1998)
Copyright (C) 1989-98 INRIA
loading initial environment
at the "-->" prompt, enter:
This simply means "execute the scilab commands from the file called
'auto.tst'". The ; means "don't echo command to the screen.
This will cause a plot window to appear. When you get the "halt"
message in the main scilab window, press "enter" to continue. You
will have to do this for about 15-20 times (press "enter" at the
"halt" message that is). When thats all done (no more "halt"
messgages), select the "File" menu from the main scilab window.
At this point, bug#1 on my Mac appears. On my mac (NetBSD-1.3), the
yellow drop down menu is as wide as the entire scilab window while on
the pmax and i386 (both NetBSD-1.3.2), the menu is only as wide as
the text of the menu. As a note, running on either pmax or i386 with
the display set back to my mac, produces the bug while running with
the display on either the pmax or i386, there is no display bug.
I'm not too concerned about this one, it just doesn't look pretty...
I tend to believe its a bug in the X stuff on my mac.
Under the "File" menu in the main scilab window, choose "File
Operations". When the file dialog box comes, select "auto.tst" with
the mouse. Now press the "Exec" button. At this point, bug#2
appears. On the mac, the program simply exits right away. On the
i386 and pmax, the commands in auto.tst are run exactly the same way
as when I simply typed
My mac workaround for now is to run scripts from the command line
instead of the pulldown menus, but the menu's really should work.
I'm hoping that the 2 bugs listed above are only on NetBSD-1.3/mac68k and
that they go away when built on a NetBSD-1.3.2/mac68k machine.
Anyway, I'd appreciate any feedback before I submit this package. Its so
much bigger than anything I've packaged, there is a lot of room for me to
have missed something (especially with the linking stuff on some of the
platforms I've never messed with and don't know anything about).