New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pari fails to build on AIX #9990
Comments
comment:1
The extension The extension |
comment:2
Replying to @nexttime:
I wonder what the logic is for having both shared and static? I'd rather use the right extension on the right platform. I can discuss this with the Pari developers after getting clarification from some AIX experts on comp.unix.aix - they are a very helpful bunch. |
comment:3
Replying to @sagetrac-drkirkby:
For an executable that doesn't make that much sense, but And since PARI is (also) a library, it's pretty normal to build both. (Every proper library does this, at least if you |
comment:4
I've reported this upstream:
The bug reported included a log against the latest snapshot, which shows the same problem. Dave |
Upstream: Reported upstream. Little or no feedback. |
This comment has been minimized.
This comment has been minimized.
comment:7
I got this from Bill Allombert, a Pari developer:
That solves the problem - Pari then installs ok. I will create a patch. Dave |
comment:8
Replying to @sagetrac-drkirkby:
Did you only try to compile PARI/GP stand-alone, or also within Sage? Because I don't know whether it is possible to compile Sage statically. |
comment:9
Replying to @sagetrac-drkirkby:
If you do create a new pari spkg, please base it on the one at #9959: http://sage.math.washington.edu/home/jdemeyer/spkg/pari-2.4.3.svn-12577.p9.spkg |
comment:10
Replying to @jdemeyer:
Only outside of Sage at this point. Some of the ATLAS libraries are only stati, but Sage links to them ok. Since Sage will not build fully on AIX, it's hard to know for sure what problems might present themselves at a later date, but I doubt a static Pari library would be one of them. I later found out that it appears that on AIX the .a files are archives which can contain both static and shared objects. But using .so as an extension for shared objects is not a problem in itself. |
This comment has been minimized.
This comment has been minimized.
comment:12
Replying to @jdemeyer:
No problem. This should be a very low-risk patch, as it will be specific to AIX. But I have some other more important things to do now, so it will have to wait. Dave |
Changed upstream from Reported upstream. Little or no feedback. to Reported upstream. No feedback yet. |
comment:14
"No feedback yet." is totally not correct here. Most applicable would be "Developers sort-of acknowledge the bug but don't really care enough to fix it" |
Changed upstream from Reported upstream. No feedback yet. to Reported upstream. Developers acknowledge bug. |
comment:15
I believe "-Wl,-r" is incorrect, "-Wl,-brtl" may have to be added for the linking to happen, as in the linking of binary. |
comment:16
And with a little bit of fiddling in Oaix-ppc/Makefile
The warning is about the fact that bcopy is in both libreadline and libc. Probably because I have both libreadline.a and .so at the moment and .a would have chosen over .so. That's minor compared to some other stuff like the fact that it segfault:
For info that's on a power7 machine running aix-6.1(TL6) with gcc-4.6.3 [compiled on another LPAR on the same machine running TL7]. |
comment:21
Closing as obsolete. |
The Pari svn snapshot 12577 is failing to build properly on AIX. If the Pari developers would like access to AIX hardware (I don't think they have it), then I can provide access to my personal RS/6000
Reported upstream: http://pari.math.u-bordeaux.fr/cgi-bin/bugreport.cgi?bug=1102
Hardware and software
== The Problem ==
A full build log is attached, but here is the end which show the failure.
Upstream: Reported upstream. Developers acknowledge bug.
CC: @jdemeyer
Component: porting: AIX or HP-UX
Issue created by migration from https://trac.sagemath.org/ticket/9990
The text was updated successfully, but these errors were encountered: