axiom-developer
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-developer] Axiom crashing in Zope-Plone


From: Camm Maguire
Subject: Re: [Axiom-developer] Axiom crashing in Zope-Plone
Date: 01 Oct 2004 11:43:58 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2

Greetings!  Bill your summary is wonderful!  Thanks!

"Page, Bill" <address@hidden> writes:

> Cam,
> 
> The original report from Hans:
> 
>   http://lists.gnu.org/archive/html/axiom-developer/2004-09/msg00056.html
> 
> referred to what later was discovered to be a seg fault
> (EXITSTATUS:139 from Python 2.2 popen3) generated by Axiom
> when run as a sub-process (popen3). Hans later reported that
> the failure occured while running a debian binary version
> of Axiom on an Intel i386 box.
> 

My suspicion lies in python's pipe code, possibly in conjunction with
GCL readline.  Can a segault be reproduced with axiom intel
standalone?  Alternatively, can someone please try to reproduce
setting the TERM environment variable to "dumb"?  Might need to check
strace -f on the process to make sure Python does not setup the pipe
environment as a readline capable terminal (e.g. vt100).  If one can
get to axiom at all over the pipe, try doing )lisp (si::readline-off)
at the beginning, or putting this into some axiom rc startup file or
something. 

Please feel free to report all Debian binary bugs to the Debian BTS
via 'reportbug'.

> In
> 
>   http://lists.gnu.org/archive/html/axiom-developer/2004-09/msg00059.html
> 
> Bob McElrath reported that he obtained the same failure
> when trying to run a debian binary version of Axiom as a
> sub-process on an Alpha debian system. It also turned
> out that Bob observed the seg fault even when Axiom was
> not run as a sub-process.
> 

(separate email)

> In the message
> 
>   http://lists.gnu.org/archive/html/axiom-developer/2004-09/msg00062.html
> 
> and
> 
>   http://lists.gnu.org/archive/html/axiom-developer/2004-09/msg00064.html
> 
> Hans reported success on the same hardware platform and
> debian version but with Axiom re-compiled from source.
> 
> In
> 
>   http://lists.gnu.org/archive/html/axiom-developer/2004-09/msg00065.html
> 
> Bob McElrath supplied some details of the debian and
> axiom installation on his Alpha system which was still
> failing with a seg fault.
> 
> Your suggested possible cause being "libc et.al dynamic
> dependencies" in the case of Bob's system seems plausible
> to me. I expect he will get back to you about this soon.
> 
> ----------
> 
> Since Hans' problem was solved with a re-compile of Axiom
> from source (and a different, older? version of gcl) on a
> debian Intel platform, I would worry that this might
> indicated a memory management problem which is more severe
> when Axiom is run as a sub-process - perhaps with limited
> physical memory since Python + Zope + LaTeX + +Ghostscript
> + Axiom is a pretty heavy combination.
> 
> Hans said that he would be busy with his "day job" for
> the next week but that he intended to try to build the
> MathAction configuration again from scratch to see he
> he gets the same problem and solution.
> 
> Thanks for looking into this!

Great!  Please keep me posted.

Take care,


> 
> Regards,
> Bill Page.
> 
> > -----Original Message-----
> > From: Camm Maguire [mailto:address@hidden
> > Sent: Tuesday, September 28, 2004 2:21 PM
> > To: Bob McElrath
> > Cc: address@hidden; Bill Page; 'Hans Peter Wuermli'
> > Subject: Re: [Axiom-developer] Axiom crashing in Zope-Plone
> > 
> > 
> > Greetings!  Haven't heard anything here -- just wondering if this
> > issue is still live.
> > 
> > Take care,
> > 
> > Camm Maguire <address@hidden> writes:
> > 
> > > Greetings!  My apologies, but I've lost the head of this 
> > thread.  I'm
> > > assuming that there is a segfault problem with debian alpha axiom
> > > (unstable).  (GCL is built in btw).  I'm also assuming the problem
> > > shows up on startup.  In this case, I cannot reproduce -- just
> > > downloaded the binaries and installed on 
> > escher.debian.org's unstable
> > > dchroot.  Of course you can only run these on Debian testing or
> > > unstable due to the libc et.al dynamic dependencies.  Is that the
> > > problem?  If any of this is close, please fill in a 
> > detailed way I can
> > > reproduce the crash, or better yet, send a copy both to me 
> > and to the
> > > Debian BTS.  'reportbug' is good for this.
> > > 
> > > For the package to build, btw, the compiled axiom must successfully
> > > run the full input test suite, so the very existence of the alpha
> > > package means that this was done.  You can find a log on
> > > buildd.debian.org if interested.
> > > 
> > > Take care,
> > > 
> > > Bob McElrath <address@hidden> writes:
> > > 
> > > > Bill Page address@hidden wrote:
> > > > > Hans,
> > > > > 
> > > > > Do I understand correctly that after re-compiling Axiom that
> > > > > you now have a fully operational LatexWiki system under debian
> > > > > that can run Axiom? If so, congratulations! I think there are
> > > > > a number of other people (including Tim Daly :) who would very
> > > > > much like to setup a similar stand alone LatexWiki/Axiom system.
> > > > 
> > > > And me!
> > > > 
> > > > But we really need to have people setting this up in a 
> > chroot jail.  I
> > > > have been reading up on this...we need a script to create 
> > the jail.
> > > > 
> > > > > I promised Tim that I would set something up for this, but I
> > > > > have not yet had the time. If you have some time to write up a
> > > > > short description of how you did it, I think it would be most
> > > > > appreciated.
> > > > 
> > > > I would like to distribute axiom support with LatexWiki.  
> > I have tied my
> > > > version numbers to ZWiki, and they will release 0.35 on 
> > 10/1.  Do you
> > > > think we could add this by then?  (or maybe a bit after 
> > to ensure 0.35
> > > > compatability)
> > > > 
> > > > Bill our darcs repos have diverged, your patches no 
> > longer apply cleanly
> > > > to latexwiki.  I have moved the functions 'runCommand' 
> > and 'log' into
> > > > util.py so that they can be used by axiom/reduce.  There 
> > are a couple
> > > > other conflicts too (I improved plone/stylesheet in 
> > latexwiki.css).  Why
> > > > do you have a font-size +2 in your latexwiki.css?  Will 
> > you have time to
> > > > merge this in the next week or two?
> > > > 
> > > > FYI, I have gotten MathML/LatexWiki working in Plone::
> > > > 
> > > >     http://mcelrath.org/Plone/ITeXTest
> > > > 
> > > > You will need my zwiki patches to do that::
> > > > 
> > > >     http://bob.mcelrath.org/darcs/zwiki
> > > > 
> > > > Hopefully I will get those into ZWiki 0.35 but I haven't 
> > been able to
> > > > get Simon's attention in the last couple of days.  ;)  As 
> > a reminder my
> > > > latexwiki repo is here::
> > > > 
> > > >     http://bob.mcelrath.org/darcs/latexwiki
> > > > 
> > > > > Did you download the binary version of Axiom that failed from
> > > > > debian unstable? If so, we should be sure to let Camm Maguire
> > > > > know since he is the architect of the debian version. What
> > > > > version of GCL are you using for the re-compile? What is the
> > > > > hardware platform that you are using?
> > > > 
> > > > Mine is a debian binary::
> > > > 
> > > >     (0)<address@hidden:~> dpkg -l axiom
> > > >     Desired=Unknown/Install/Remove/Purge/Hold
> > > >     | 
> > Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-
> > installed
> > > >     |/ Err?=(none)/Hold/Reinst-required/X=both-problems 
> > (Status,Err: uppercase=bad)
> > > >     ||/ Name              Version           Description
> > > >     
> > +++-=================-=================-======================
> > ============================
> > > >     ii  axiom             0.20040831-1      A general 
> > purpose computer algebra system: main bi
> > > >     (0)<address@hidden:~> axiom
> > > >     Segmentation fault
> > > > 
> > > > This is the alpha architecture.
> > > > 
> > > > If I understand correctly, GCL is compiled-in?  I do not 
> > have gcl itself
> > > > installed on this machine.
> > > > 
> > > > --
> > > > Cheers,
> > > > Bob McElrath [Univ. of California at Davis, Department of Physics]
> > 
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah




reply via email to

[Prev in Thread] Current Thread [Next in Thread]