Sat 02 Jul 2016 04:59:01 PM UTC, comment #11:
Yes, it downloads the packages and copies them to the $(HOST_PREFIX)/src directory for distribution.
|
Sat 02 Jul 2016 04:28:05 PM UTC, comment #10:
@jwe: Okay. Even without --enable-binary-packages the log shows that it is doing something with the of-packages beyond just downloading them.
Given that this report was about a problem with gid_t, uid_t, nlink_t that has been resolved, I will still leave this closed.
I'll re-configure with --enable-binary-packages and open up a new bug report if I can't build.
|
Sat 02 Jul 2016 04:22:31 PM UTC, comment #9:
Rik, did you use --enable-binary-packages? If not, then you weren't actually compiling any of the OF packages.
|
Sat 02 Jul 2016 04:20:43 PM UTC, comment #8:
I just built MXE-Octave completely from scratch for a 32-bit Windows system and didn't encounter any problems with any of the OF packages.
I'm going to close this report. If there are any specific packages that fail to build later they can get a new bug report.
|
Fri 01 Jul 2016 07:51:15 PM UTC, comment #7:
On my win32 build I didn't have any issues with control.
I do get failures on:
sparserb - bug #48335
Nan
odepkg (although forge dev sources work)
zeromq (forge dev sources work)
fl-core (forge dev sources work)
lssa
communications
|
Fri 01 Jul 2016 01:04:40 PM UTC, comment #6:
No more issues after first cleaning up mxe-octave before building a new dev Octave.
I noted that of-control didn't get built in 32-bit mxe-octave
Could be a hick-up - in 64-bit mxe-octave octave it built fine.
|
Fri 01 Jul 2016 11:49:36 AM UTC, comment #5:
Pushed http://hg.savannah.gnu.org/hgweb/octave/rev/2d8e9bdf5683
|
Fri 01 Jul 2016 07:53:38 AM UTC, comment #4:
64.
I didn't try 32-bit build (could do that as well later today, see below)
I've re-cloned and deleted the entire build dir (in Linux). Should be OK.
But ... maybe I'd better also clean up mxe-octave from all Octave remnants the way I described it some days ago for Rik in the maintainers ML.
So, hang on .... (later today, after work, I'll have a go)
(New bug report needed, or hijack this one and just change title?)
|
Thu 30 Jun 2016 10:44:57 PM UTC, comment #3:
Is that 64 bit or 32 bit?
|
Thu 30 Jun 2016 09:56:59 PM UTC, comment #2:
Issue confirmed, and confirmed that the patch works.
However I now get other issues on the Windows side:
- colormap errors and deprecated PKG_ADD messages during postprocessing the installation on Windows. That phase takes a looong time now (the cmd32 window is empty for a minute or so, then prints these messages and a few seconds after that closes).
- The GUI won't start; I have to kill the octave-gui process through Task manager
The CLI does start but with errors:
and the help system doesn't work:
... although maybe ,aybe that could be related to the trick Lachlan showed in bug #48172 (see comment #44 there) to build dev Octave on Linux.
An earlier build using that trick did have a functional help system (and its GUI started up fine).
|
Thu 30 Jun 2016 04:56:44 PM UTC, comment #1:
Patch attached
(file #37635)
|
Thu 30 Jun 2016 04:32:24 PM UTC, original submission:
Using f471f513ffd0 also define gid_t, uid_t, and nlink_t in octave-config.h (bug #48332)
octave fails to compile:
Due to changes from 6bce4d23af6b eliminate OCTAVE_USE_WINDOWS_API and OCTAVE_USE_OS_X_API from header files
|