Robert Bradshaw wrote:
> On Nov 3, 2009, at 5:28 PM, Dr. David Kirkby wrote:
> 
>> If one tries to build Sage on OpenSolaris, python will not build,  
>> complaining
>> about a missing md5 module.
>>
>>
>>   File "<string>", line 1, in <module>
>>   File "/export/home/drkirkby/sage-4.2/local/lib/python/hashlib.py",  
>> line 136,
>> in <module>
>>     md5 = __get_builtin_constructor('md5')
>>   File "/export/home/drkirkby/sage-4.2/local/lib/python/hashlib.py",  
>> line 63,
>> in __get_builtin_constructor
>>     import _md5
>> ImportError: No module named _md5
>>
>> real    2m15.850s
>> user    1m46.382s
>> sys    0m19.649s
>> sage: An error occurred while installing python-2.6.2.p4
>>
>>
>> William said the other day the way around this is to build the  
>> OpenSSL libraries
>> first. Then I looked and see that the OpenSSL libraries were at one  
>> time
>> included in Sage, but were removed since they are not GPL. This was  
>> done in
>> ticket #478
>>
>> http://trac.sagemath.org/sage_trac/ticket/478
>>
>> So I'm not sure what to do here. Clearly I could install the OpenSSL  
>> libraries,
>> and will no doubt do that so progress can be made. But this seems  
>> quite a
>> fundamental problem. There seems two choices
>>
>> * Include OpenSSL with Sage, which we can't do as its not GPL
>> * Tell people to build that themselves, which is still meaning
>>
>> BTW, I was hoping to get my changes for the code which checks the  
>> prerequisites
>> for Sage (version 0.5)  reviewed.
>>
>> http://trac.sagemath.org/sage_trac/ticket/7352
>>
>> Should OpenSSL library support be considered a prerequiste for Sage?
> 
> I don't think so, at least not yet. If Sage builds fine and passes all  
> doctests except for this one detail, then perhaps we should bring it  
> up again at that point, but it really sounds like a bug in the  
> OpenSolaris linker (not being able to build gnutls) and adding a  
> prerequisite seems like a heavy-handed fix.
> 
> - Robert

gnutls builds fine with only one set of libraries. I do not know if that is a 
linker bug, or a bug in the way the linker is called, which uses libtool.

According to this very old trac ticket

http://trac.sagemath.org/sage_trac/ticket/478

"Unfortunately, gnutls's presence isn't enough for Python to build the "md5" 
module, and SAGE needs that module. "

But looking in a bit more detail, the openssl library exists on the Solaris 
system, version 0.9.8. So I'm not sure why python would not build. William said 
to build that.

Note, the OpenSolaris system here is a Xeon, not a SPARC. (Solaris 10 will run 
on x86 and OpenSolaris will run on SPARC). I just do not happen to have any of 
those combos.

r...@hawk:~# uname -a
SunOS hawk 5.11 snv_111b i86pc i386 i86pc
r...@hawk:~# ls -l /lib/libssl.so
lrwxrwxrwx   1 root     root          17 Oct 26 17:17 /lib/libssl.so -> 
./libssl.so.0.9.8



--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---

Reply via email to