Michael Goffioul wrote: > On Wed, Jul 13, 2011 at 4:15 PM, Philip Nienhuis<pr.nienh...@hccnet.nl> > wrote: >> Hi Michael: >> >> (octave-dev@lists.sourceforge.net cc'ed) >> >> Michael Goffioul wrote: >>> >>> On Tue, Jul 12, 2011 at 7:30 PM, Philip Nienhuis<pr.nienh...@hccnet.nl> >>> wrote: >>>> >>>> Michael Goffioul wrote: >>>>> >>>>> The backtrace is weird, it looks like polymorphism isn't use, as in >>>>> octave_base_value::subsref >>>>> is called instead of the overridden octave_com_object::subsref. I >>>>> assume you recompiled the >>>>> Windows package as well. >>>> >>>> No - how does one do that? >>>> >>>> (I did figure it may be needed but didn't know how to get it together, so >>>> I >>>> skipped that step.) >>>> >>>> I'll try some more backtracing earlier on in the code as you suggested, >>>> but >>>> I have to sneak that in between other "duties". >>> >>> Grab the package sources from sourceforget and try "pkg install...". >> >> Yeah, it occurred to me this morning that installing an octave-forge package >> might invoke the very compiler flags used to build Octave itself. So if >> debug was enabled for Octave, it'll apply to packages too. >> Or is this a wrong supposition? >> (BTW: This may be something for the wiki. This info may be obvious for the >> gurus, but is nowhere to be found.) >> >> Now, I have done "pkg install -forge windows" from the debug octave build. >> So the backtrace I posted was from a "debug" windows package after all. >> >> Hopefully I 'll have a chance to dig further somewhere the next days. >> >> Philip >> > > Hi Philip, > > After being inactive for a while, I recently recompiled octave and a bunch > of forge packages. So I could debug the issue you reported and found out that > the code I was looking at (SVN) is not the one that is downloaded from the > octave-forge download pages. The SVN code contains some changes that > should fix the issue. So it's just a matter of releasing a new version of the > Windows package.
This does ring a bell - I already stated a few posts back that I recognized the symptoms from -maybe- two years back, or longer. Given your explanation above it is intruiging that Benjamin managed to compile it for the 3.2.4-MinGW binary. I also couldn't track down the changes you mention. I d/led a GNU tarball, copied all files in it over into an unzipped windows-1.0.8.tar.gz package, ran ./autogen.sh, tar.gz-ed it again and tried to install it in Octave-MinGW-3.4.2, but I got tons of warnings and several compile errors. But admittedly the 3.4.2 development environment I have with me on my laptop (abroad) may not be perfect. Did you succeed in installing it yourself? > Given my limited available time, It can take me a while before I release a new > version of the package, so I'd suggest you do it instead (if you have the > time, > of course). Sure, if I can get the package to compile myself. And it may take a while as I haven't got much time in September. Better to first try to get it to compile & work OK with Octave 3.4.x & 3.5.+ (MinGW; you still run MSVC, don't you?) before it's released. Hopefully somewhere next week I can tell you how far I got. Thanks, Philip ------------------------------------------------------------------------------ Special Offer -- Download ArcSight Logger for FREE! Finally, a world-class log management solution at an even better price-free! And you'll get a free "Love Thy Logs" t-shirt when you download Logger. Secure your free ArcSight Logger TODAY! http://p.sf.net/sfu/arcsisghtdev2dev _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
