Opened 11 years ago
Closed 10 years ago
#9264 closed defect (duplicate)
Apply ALL relevent fixes to ECL 10.4.1
Reported by: | drkirkby | Owned by: | tbd |
---|---|---|---|
Priority: | major | Milestone: | sage-duplicate/invalid/wontfix |
Component: | packages: standard | Keywords: | |
Cc: | was, jhpalmieri, mpatel, leif, jsp, jason, vbraun, jdemeyer | Merged in: | |
Authors: | Reviewers: | Volker Braun | |
Report Upstream: | N/A | Work issues: | |
Branch: | Commit: | ||
Dependencies: | Stopgaps: |
Description (last modified by )
There are a number of tickets active at the minute with patches to ECL
none have been merged, but the patches are not consistent across all or them.
- The option --with-dffi=no from #8089 has not been been missed which is very important.
- Some other less important changes (such as removing an inaccurate, but innocuous mention of building a 32-bit version) are removed.
- There is a 10.4.1.p0 package, despite no 10.4.1 has actually been merged.
The aim of this ticket is to update to the 10.4.1 source code, whilst ensuring all the relevant patches are applied.
Important
On #8951 Jason says "As a reminder here, this (i.e. #8951) should *only* be merged simultaneously with the new maxima at #8731."
Assuming Jason is correct, that gives us a problem.
The Maxima update seems to have stalled (there are no updates on the ticket for 3 weeks). If updating Maxima to 5.21.1 and ECL 10.4.1 need to be done together, it will probably be best to create a ecl-10.2.1.p0, which has all the small changes, and actually leave updating to the latest ECL code until it can be better coordinated with the Maxima update.
At the moment, it seems several small but important changes to the ECL installation are being held up for a non-trivial set of changes needed if Mamima is updated.
Dave
Change History (22)
comment:1 Changed 11 years ago by
- Cc jason added
- Description modified (diff)
comment:2 Changed 11 years ago by
comment:3 Changed 11 years ago by
- Status changed from new to needs_review
I've created a new ECL package
http://boxen.math.washington.edu/home/kirkby/revised-patches/ecl-10.4.1.spkg
which has the following fixes applied:
- Updates ECL to the latest upstream, 10.4.1
- Applies a fix to stop ECL building in parallel, which is ticket #9187. This already has positive review.
- Applies the fix from #8089 to disable the dynamic foreign function interface as advised by the ECL developer. This disables the use of assembly code on OpenSolaris x64, but no other platform.
- Removes files from /tmp/ECL* as soon as they are no longer needed, as failure to do so can present problems on multi-user systems.
All changes have been checked with #8645, which has a change to Maxima which allows Maxima to build with the latest ECL. The patch on #8645 does not update Maxima to the latest version, as that creates problems which are non-trivial to solve. It does however fix a problem with Maxima not installing the library properly, and allows Maxima to build with the latest ECL>
As such, the attached ticket:
- Allows Maxima to build the library properly whilst still working with the latest ECL.
- Remove unwanted ECL tmp files.
- Allows spkgs to build in parallel
- Allow ECL to build on OpenSolaris x64.
Here are the test results:
Testing on Solaris 10 SPARC in 32-bit mode.
- Sun Blade 1000
- 2 x 900 MHz UltraSPARC III+ CPUs
- 2 GB RAM
- Solaris 10 03/2005 (first release of Solaris 10)
- gcc 4.4.3 (uses Sun linker and assembler)
- Sage 4.4.4.alpha1
done for i in Copyright LGPL; do \ /usr/bin/ginstall -c -m 644 /export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/ecl-10.4.1/src/src/../$i /export/home/drkirkby/sage-4.4.4.alpha1/local/lib/; \ done /bin/sh /export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/ecl-10.4.1/src/src/gc/mkinstalldirs /export/home/drkirkby/sage-4.4.4.alpha1/local/share/man/man1 for i in doc/ecl.man doc/ecl-config.man; do \ /usr/bin/ginstall -c -m 644 $i /export/home/drkirkby/sage-4.4.4.alpha1/local/share/man/man1/ecl.1; \ done make[1]: Leaving directory `/export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/ecl-10.4.1/src/build' real 1m26.075s user 1m17.655s sys 0m7.531s Successfully installed ecl-10.4.1
Here is the Maxima 5.20.1.p1 installation from #8645
installing Maxima library as /export/home/drkirkby/sage-4.4.4.alpha1/local/lib/ecl//maxima.fas real 22m32.432s user 19m9.372s sys 2m51.674s Successfully installed maxima-5.20.1.p1 Now cleaning up tmp files. rm: Cannot remove any directory in the path of the current working directory /export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/maxima-5.20.1.p1 Making Sage/Python scripts relocatable... Making script relocatable Finished installing maxima-5.20.1.p1.spkg drkirkby@redstart:~/sage-4.4.4.alpha1$
Testing on OpenSolaris x64 06/2009 64-bit mode.
- Sun Ultra 27
- 2 x 3.33 GHz quad core Intel Xeon MHz
- 12 GB RAM
- OpenSolaris 06/2009 (Last release of OpenSolaris, updated to build 134)
- gcc 4.4.4 (uses Sun linker and GNU assembler)
- Sage 4.4.4.alpha1
/bin/sh /export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/ecl-10.4.1/src/src/gc/mkinstalldirs /export/home/drkirkby/sage-4.4.4.alpha1/local/share/man/man1 for i in doc/ecl.man doc/ecl-config.man; do \ /usr/bin/ginstall -c -m 644 $i /export/home/drkirkby/sage-4.4.4.alpha1/local/share/man/man1/ecl.1; \ done make[1]: Leaving directory `/export/home/drkirkby/sage-4.4.4.alpha1/spkg/build/ecl-10.4.1/src/build' real 1m26.075s user 1m17.655s sys 0m7.531s Successfully installed ecl-10.4.1
Maxima 5.20.1.p1 will not build on this platform, but such a problem is not related to this ticket, and appears it might be an incorrect flag given in the source (-Wl,-G instead of -shared). This is documented at #9099
Testing on sage.math (Linux), 64-bit.
n/man1/ecl.1; \ done make[1]: Leaving directory `/home/kirkby/sage-4.4.3/spkg/build/ecl-10.4.1/src/build' real 2m16.904s user 1m44.990s sys 0m14.980s Successfully installed ecl-10.4.1 Now cleaning up tmp files. Making Sage/Python scripts relocatable... Making script relocatable Finished installing ecl-10.4.1.spkg
We can also see Maxima, and the library are ok now.
;;; installing Maxima library as /home/kirkby/sage-4.4.3/local/lib/ecl//maxima.fas real 4m41.594s user 3m14.160s sys 0m46.000s Successfully installed maxima-5.20.1.p1 Now cleaning up tmp files. Making Sage/Python scripts relocatable... Making script relocatable Finished installing maxima-5.20.1.p1.spkg
Note to Release manager
There are many tickets active at the minute for updates to Maxima and ECL. Once positively, this should be integrated with Maxima at #8645. At the time of writing, the use of the latest Maxima will cause problems with doctest failures, as Maxima's output is changed from the current version.
comment:4 Changed 11 years ago by
- Reviewers set to Nils Bruin
- Status changed from needs_review to positive_review
I have tested this ticket by building sage 4.4.3 with the following spkgs substituted:
http://boxen.math.washington.edu/home/kirkby/revised-patches/ecl-10.4.1.spkg
http://sage.math.washington.edu/home/nbruin/maxima-5.20.1.p1.spkg
All doctests pass on sage.math.washington.edu
. Furthermore, the ecl-10.4.1 package looks like it does what it's supposed to. When I diff it with http://sage.math.washington.edu/home/nbruin/ecl-10.4.1.spkg there are some unexpected files and differences in binary files in the .hg
subdir, but otherwise the packages agree.
Merging this ticket should be a good stepping stone towards #8731 (assuming the issues there haven't been resolved by the time the next release merges are made). Merging this ticket would also resolve #8645, as well as all ECL related ticket referred to.
comment:5 Changed 11 years ago by
I just noticed a small inaccuracy in the description of the hardware used to test this on OpenSolaris. It does not affect the validity of the ticket, but it is sensible to correct the error.
The section marked Testing on OpenSolaris x64 06/2009 64-bit mode. states the processors are:
- 2 x 3.33 GHz quad core Intel Xeon MHz
which is incorrect. Unfortunately, the Sun Ultra 27 does not support more than one CPU. A more accurate description of the hardware/software used to test this ECL package on OpenSolaris is:
Testing on OpenSolaris x64 06/2009 64-bit mode.
- Sun Ultra 27 workstation
- 1 x 3.33 GHz 64-bit Intel Xeon W3580 CPU (4 cores, 8 threads, 8 MB cache).
- 12 GB RAM
- OpenSolaris 06/2009 (Last release of OpenSolaris, updated to build 134)
- gcc 4.4.4 (uses Sun linker and GNU assembler)
- Sage 4.4.4.alpha1
Dave
comment:6 Changed 11 years ago by
- Merged in set to sage-4.5.alpha0
- Resolution set to fixed
- Status changed from positive_review to closed
comment:7 Changed 11 years ago by
comment:8 Changed 11 years ago by
- Merged in sage-4.5.alpha0 deleted
- Resolution fixed deleted
- Status changed from closed to new
comment:9 Changed 10 years ago by
- Cc vbraun added
ecl-10.2.1.p3 does not compile on Fedora 14:
libeclmin.a(sequence.o): In function `cl_sublis': /home/vbraun/Code/ecl/src/c/sequence.d:106: multiple definition of `cl_sublis' libeclmin.a(list.o):/home/vbraun/Code/ecl/src/c/list.d:778: first defined here c/all_symbols.o:(.data.rel+0x9b20): undefined reference to `cl_subseq' libeclmin.a(string.o): In function `string_trim0': /home/vbraun/Code/ecl/src/c/string.d:756: undefined reference to `cl_subseq' libeclmin.a(sequence.o): In function `cl_copy_seq': tmp.c:(.text+0x4c0): undefined reference to `cl_subseq' libeclmin.a(pathname.o): In function `make_one': /home/vbraun/Code/ecl/src/c/pathname.d:234: undefined reference to `cl_subseq' /home/vbraun/Code/ecl/src/c/pathname.d:234: undefined reference to `cl_subseq' collect2: ld returned 1 exit status make[1]: *** [ecl_min] Error 1
The current ecl-10.4.1 compiles fine. Given that Fedora 14 is scheduled to be released next Monday, is there any chance we can update ecl? Ideally in the next Sage (4.6.1) release...
comment:10 Changed 10 years ago by
You should open a new trac ticket for this.
Dave
comment:11 Changed 10 years ago by
I've created a new trac ticket for you - see #10185.
comment:12 Changed 10 years ago by
I realise this ticket was still open, but I still think the Fedora issue should go on a ticket of its own. The package I made for this ticket is not the latest ECL, so if we do update ECL, we will probably end up using a later version of it. The version here does not have some fixes which are in the current ECL in Sage. In particular, #9917 has fixes which are not in my package referenced here.
comment:13 Changed 10 years ago by
I went through ./src/src/c/dpp.c
in your ecl-10.4.1 spkg (http://boxen.math.washington.edu/home/kirkby/revised-patches/ecl-10.4.1.spkg) and checked that all fprint()
s have the right number of arguments. So upstream fixed #9917 and we don't have to patch it any more.
I removed the included gmp as discussed in #9493.
Are there any other outstanding patches? I've looked through trac and there seem to be many overlapping discussions. But I could not find any outstanding bugs.
The new ecl requires also a new maxima (5.22.1 is newest upstream release), so we have to update both at the same time. I've created #10187 for a simultaneous ecl+maxima update.
comment:14 Changed 10 years ago by
- Cc jeroen added
This ticket can be closed as we did update to ecl-10.4.1 a while ago.
comment:15 follow-up: ↓ 17 Changed 10 years ago by
- Cc jdemeyer added; jeroen removed
- Reviewers changed from Nils Bruin to Volker Braun
- Status changed from new to needs_review
In fact, now ECL is at version 11.1.1 in Sage. #9493 is still open for the GMP removal, otherwise this ticket can be closed.
comment:16 Changed 10 years ago by
- Status changed from needs_review to positive_review
comment:17 in reply to: ↑ 15 Changed 10 years ago by
comment:18 follow-ups: ↓ 19 ↓ 20 Changed 10 years ago by
Well, unpack ecl-11.1.1.p1.spkg and find out. At any rate, that is on the ticket description at #9493 still, so this ticket should be closed.
comment:19 in reply to: ↑ 18 Changed 10 years ago by
comment:20 in reply to: ↑ 18 Changed 10 years ago by
Replying to kcrisman:
Well, unpack ecl-11.1.1.p1.spkg and find out. At any rate, that is on the ticket description at #9493 still, so this ticket should be closed.
Could be the first time I ever use sage -info ...
(which ought to dump SPKG.txt
to the screen, but unfortunately currently doesn't: #11021). :D
comment:21 Changed 10 years ago by
- Milestone changed from sage-4.7.1 to sage-duplicate/invalid/wontfix
comment:22 Changed 10 years ago by
- Resolution set to duplicate
- Status changed from positive_review to closed
Your best chance is to see if you can give a positive review to #8645. The maxima spkg there:
is the same maxima that is in sage now, with minimal changes applied to let maxima build under ECL 10.4.1.
If maxima.5.20.1.p1 on ecl.10.4.1 needs work, we might as well wait until maxima.5.21 is ready to be merged, because then the problem is ecl.10.4.1.
When I tried, I couldn't get maxima to build properly under ecl.10.2 (its "ASDF" is too buggy)