octave: update to 2.9.17

[ryandesign@…]

octave 2.9.17 has been released which, according to Brent Fulgham in #12144, works better on Mac OS X Leopard. So we should update to that version.

[ryandesign@…]

I fixed the livecheck to recognize that 2.9.17 is available, in r31192.

[bfulgham@…]

Note that 2.9.17 seems to have a problem with its TeX handling. I have tried switching from teTeX to the new texLive package, but this still seems to be a problem:

/bin/sh: line 1: 11287 Segmentation fault      ./munge-texi -d ../../src/DOCSTRINGS -d ../../scripts/DOCSTRINGS < arith.txi > arith.texi-t
make[3]: *** [arith.texi] Error 139
make[2]: *** [interpreter] Error 2
make[1]: *** [doc] Error 2
make: *** [all] Error 2
Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_Users_brent_Projects_macports-trunk_dports_math_octave/work/octave-2.9.17" && make all " returned error 2
Command output: ls: *.cc: No such file or directory
make -C time all

[ryandesign@…]

See #13686 for a problem occurring with 2.9.15 on Leopard. (Don't know if 2.9.17 resolves it.)

[bcbarnes@…]

any news? Octave 2.9.18, 2.9.19, and 3.0.0 are now available. I would also like it if Texlive was a dependency (instead of Tetex), but do not know if the newer versions help this or not.

[waterson@…]

Well, this is a total kludge, but...

If you rebuild munge-texi using "g++" instead of "/opt/local/bin/g++-mp-4.2", you can restart the build and things seem to go along just fine. "munge-texi" appears to simply be a C++ utility program that preprocesses doc files. (It probably be should written in Perl -- it's 2008, folks.) So this hack should get things rolling just fine -- maybe some makefile trickery is all that's needed here to use the "stock" compiler.

The real problem seems to be related to the use of stl::map in munge-texi, and I'd guess that either g++-mp-4.2 is compiling against the wrong headers, or it's linking against the wrong libraries. The latter seems more likely, given that the stack trace mumbles about bad stack frames and gdb seems hopelessly confused by the g++-mp-4.2-compiled binary. (Plus, I know that gcc's ABI has changed over time.)

Anyway, hope that helps.

And sadly, the story doesn't have a happy ending, since the resulting octave executable dumps core on startup...I'll look a little bit more...

[waterson@…]

Sigh, I probably should've commented on #13686 -- let me go do that.

[alakazam@…]

I worked a bit on updating octave to version 3.0.0 this weekend. I will not have time until next weekend to continue working on this, so I am uploading a partially updated patch file so as to log my progress and enable people wishing to help out to contribute.

Should we open a new ticket, and close this one, or update the description, since we will probably be going directly to version 3.0.0 instead of 2.9.17 ?

The Portfile I am uploading updates version numbers and checksums for version 3.0.0, and improves whitespace formatting.

Compiling without the configure phase patches does not work because of lcamd. The configure phase patches need to be updated for octave's new configure script in order to fix this.

The livecheck regexp probably no longer works, and was based on the wording of the release announcements, which does not seem very robust (as illustrated by its failing due to the formulation of the latest news on the octave homepage).

[alakazam@…]

Portfile for octave-3.0.0

[marcuscalhounlopez@…]

In #13686, there is already a proposed Portfile for version 3.0.0.
If g95 is used as the Fortran compiler, it passes almost all of the tests (at least on Leopard).
The only failures are because the errors are larger than the tolerances (but not by much).
For an unknown reason, using gcc42 or gcc43 as the Fortan compiler causes problems.

Perhaps these two tickets should be combined?

[alakazam@…]

Humh, well, I'm quite sorry I didn't see that before. It seems much more complete than my awkward attempt, and I will try compiling it immediately. Since I don't have a macports gcc4 installation however, I think it will require a significant number of hours.

Ryan, could you tell us how best to merge / create a new ticket ? The two tickets are really centered around the problem of upgrading to Octave 3.0.0, I think, and having two tickets might induce more confusion than good.

[ryandesign@…]

Let's just move the discussion over to #13686. I'll set this ticket to "wontfix" since we won't update the port to 2.9.17; we'll just update it straight to 3.0.0, for which more work has already been done in #13686.

[anonymous]

Milestone Port Updates deleted