Octave 2.0.5 under gnu-win32 (aka cygwin32)?

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Octave 2.0.5 under gnu-win32 (aka cygwin32)?

Mumit Khan-2
Has anybody actually built octave-2.0.5 under gnu-win32 b17.1 and gotten
past Seg faults at startup (well, a bit into startup, in some of the
standard string class methods)?

I'm testing to see if my port of g77-0.5.20 for gnu-win32 is good enough
to build octave. Currently g77 can compile everything *but* one file,
linpack/zgedi.f, but can't test the results until I get through the initial
hurdle of getting octave to run.

Thanks
Mumit

Reply | Threaded
Open this post in threaded view
|

Octave 2.0.5 under gnu-win32 (aka cygwin32)?

John W. Eaton-6
On 16-Mar-1997, Mumit Khan <[hidden email]> wrote:

| Has anybody actually built octave-2.0.5 under gnu-win32 b17.1 and gotten
| past Seg faults at startup (well, a bit into startup, in some of the
| standard string class methods)?

The disk I was using for Windoze 95 failed, so I haven't been able to
do any more work with the port since somewhere around Octave 1.93 or
1.94, I think.  I did have it working to the point of printing the
prompt and at least inverting a matrix or two.

Does gdb provide any useful information about where the crash is
happening?

Please send any additional information about the crashes that you are
seeing to [hidden email].

Thanks,

jwe


Reply | Threaded
Open this post in threaded view
|

WORKS [Re: Octave 2.0.5 under gnu-win32 (aka cygwin32)?]

Mumit Khan-2
"John W. Eaton" <[hidden email]> writes:

> On 16-Mar-1997, Mumit Khan <[hidden email]> wrote:
>
> | Has anybody actually built octave-2.0.5 under gnu-win32 b17.1 and gotten
> | past Seg faults at startup (well, a bit into startup, in some of the
> | standard string class methods)?
>
> The disk I was using for Windoze 95 failed, so I haven't been able to
> do any more work with the port since somewhere around Octave 1.93 or
> 1.94, I think.  I did have it working to the point of printing the
> prompt and at least inverting a matrix or two.

Couldn't find a copy of pre-2.0 anywhere, so couldn't really test if there
is any difference or not.

>
> Does gdb provide any useful information about where the crash is
> happening?

No, other than spurious crashes. I have full debugging info in all
of cygwin32, and even that doesn't help.

>
> Please send any additional information about the crashes that you are
> seeing to [hidden email].
>

Well, found the problem! Any CYGWIN32 program that is linked with
-lkernel32 explicitly on the link line WILL FAIL mysteriously!!!!

Octave configure extracts all the CXXLIBS and FLIBS and then explicitly
links them, and this causes weird bugs in both gnu-win32 and ming32.

Octave now starts up, inverts a few matrices ... any standard test suite
that can run w/out dejagnu installed?

Regards,
Mumit -- [hidden email]
http://www.xraylith.wisc.edu/~khan/

Mumit

Reply | Threaded
Open this post in threaded view
|

WORKS [Re: Octave 2.0.5 under gnu-win32 (aka cygwin32)?]

John W. Eaton-6
On 27-Mar-1997, Mumit Khan <[hidden email]> wrote:

| Octave now starts up, inverts a few matrices ... any standard test suite
| that can run w/out dejagnu installed?

You can run all the .m files in the tests directory, but you will
have a hard time making much sense of the results without DejaGNU.

Except for the ones that result in parse errors, I think the tests
could probably be rewritten to be useful without DejaGNU.  It would be
a fair amount of tedious work, but it would not be too difficult.  I
don't plan to do it myself, but anyone else would like to work on this
project, please send me mail.

Thanks,

jwe