bug python2.3+zipimport+ubuntu 10.04 amd_64

[Robin Becker]

OK it seems there's a bug in zipimport.c that was fixed in 2.4. I copied across 
the 2.4 version and after removing a single new error check the rebuilt python 
2.3 works fine. There was supposed to be a patch here

http://cvs.sourceforge.net/viewcvs.py/python/python/dist/src/Modules/zipimport.c?r1=1.16&r2=1.17

but it's a dead link.
-- 
Robin Becker
On 11/11/2011 16:52, Robin Becker wrote:
> I'm trying to run some aged software on a new machine which runs
>
> Linux app2.reportlab.com 2.6.32-30-generic #59-Ubuntu SMP Tue Mar 1 21:30:46 UTC
> 2011 x86_64 GNU/Linux Ubuntu 10.04.2 LTS
>
> The software is required to use python 2.3 so first I've had some problems
> building python itself. Apparently to get rid of segfaults during the build we
> need the configure args to include
>
> BASECFLAGS=-U_FORTIFY_SOURCE
>
> that certainly fixes the python build errors.
>
> Now when we come to run the aged application we get this
>
> $ ./xxx.cgi
> Traceback (most recent call last):
> File "/home/rptlab/website/develop.reportlab.com/cgi-bin/xxx.cgi", line 7, in ?
> from fastwebapp import FastWebApp
> OverflowError: signed integer is greater than maximum
>
>
> this appears to be related to the use of a zip import.
>
>  > sys.path.insert(0,os.path.join(os.getcwd(),"fastapp.zip"))
>  > from fastwebapp import FastWebApp
>
> if we unpack the zip and remove the path insert the application appears to run
> fine.
>
> Googling appears to indicate that others have seen this error but on a different
> hardware and trying to remove optimizations from the python build has had
> variable success.
>
> Anyone had any real success with this problem? The last version of the test
> machinery ran 32bit freebsd and we had no problems on that hardware.