#9914: Remove libraries from extension modules when they are not needed there at
build time
------------------------------+---------------------------------------------
Reporter: leif | Owner: leif
Type: defect | Status: needs_work
Priority: major | Milestone: sage-4.6
Component: build | Keywords: module_list.py PARI ImportError
newforms homspace mwrank upgrade update
Author: Leif Leonhardy | Upstream: N/A
Reviewer: | Merged:
Work_issues: |
------------------------------+---------------------------------------------
Comment(by leif):
Replying to [comment:4 GeorgSWeber]:
> Sorry, but this will certainly break on Cygwin!
Did you read the comments here or at the other ticket? I explicitly asked
if this is needed on Cygwin (twice), cc'ed Mike, and suggested to use
{{{uname_specific()}}} in that case.
> Just have a look at the Makefile for the "g0nntl" library (in the eclib
spkg, see the file /src/g0n/Makefile, line 41):
> [[BR]]
> NTLLFLAGS = -L$(NTLLIBDIR) -lntl -lgmp -lpari
Bad example, since you can safely remove PARI there (too; and also from
{{{NTLLFLAGS}}} in all other Makefiles); NTL uses GMP, but not PARI. (Only
{{{procs/parifact.*}}} uses PARI, so PARI has to be linked to all
executables and shared libraries that contain / use ''that'', e.g.
{{{libjcntl.{so,dylib,dll}}}}, which in turn is linked to
{{{libg0nntl.*}}}.)
(Note that "pari" is '''not''' listed in the {{{libraries}}} of
{{{sage.libs.cremona.mat}}}, though "jcntl" is, and "g0nntl", too.)
Even if e.g. {{{libg0nntl.so}}} directly depended on PARI, none of the
extension modules whose {{{libraries}}} I've changed does.
There's a difference between dynamically and statically linking btw.
> I've changed the title of the ticket somewhat, to reflect this, and
added a line to the description.
By ''"removing ... from the extension modules"'' I meant removing the in
fact (locally) unused libraries from the dynamic (dependencies) section,
i.e. for example the {{{NEEDED}}} entries in ELF.
> If this ticket is meant to fix some OS X specific issue (see the main
ticket at trac #9896), I'd propose to use the "uname_specific" feature in
such a way, that the change to the module_list.py affects Darwin --- and
only Darwin.
Again, as mentioned in my first comment here, and also at #9896, this is a
''general'' problem which is unrelated to Darwin as the OS.
> Ensuring that the change is visible on only as few system as possible,
as few systems as possible are affected of a possible breakage by such a
change.
The opposite is closer to what is needed: if at all, PARI has to be listed
in {{{libraries}}} ''only on Cygwin'' (or MS Windows) if you're right (but
see above); I'm not sure if it would be required on older Darwins as well.
So I'll update the patch to use {{{uname_specific()}}} (if that makes you
happy ;-) - I'd really like to give it a try ''as is'' on Cygwin), which
then should be tested (also) on MacOS X 10.4 (Tiger).
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/9914#comment:5>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
