Menu
▾
▴
octave-dev
|
From: Quentin S. <qsp...@ie...> - 2006-06-14 14:07:53
|
Now that 2.9.6 is out and octave-forge CVS is at least somewhat synchronized with the state of octave, I'd like to suggest a new snapshot release. The last release is now quite far out of date with respect to 2.9.x development. Quentin |
|
From: <so...@ha...> - 2007-05-22 08:32:10
|
Hi, Since 2.9.11 was just released I was wondering if we should do a new release of Octave-Forge? Until we have a system where each package maintainer can release when it suits him/her, shouldn't we just release Octave-Forge whenever there is a new Octave release? Søren |
|
From: David B. <Dav...@mo...> - 2007-05-22 08:41:31
|
Søren Hauberg wrote: > Hi, > Since 2.9.11 was just released I was wondering if we should do a new > release of Octave-Forge? Until we have a system where each package > maintainer can release when it suits him/her, shouldn't we just release > Octave-Forge whenever there is a new Octave release? > > Søren > I'd thought of that and have been working on the "make -C packages compare" target of octave-forge. It now works correctly and the MD5 sums of unmodified packages are respected, and the files to upload are placed in the package directories. It also flags versioning issues (ie modified packages without modifying the version number). I'd suggest a release of octave-forge in the next few days.. I'll be using release-forge this time as the sourceforge FRS webpages are horrible and the last release took me many hours to get right. We should also do some package testing before the upload (at least run the "make check" target). If others want to run "./configure; make all; make check" on different platforms that would help.. However, I would suggest keeping the supported version of Octave at 2.9.10 as there has been little that has changed that should affect the packaging itself.. In any case given the way the webpages are connected to the file releases and the limitations of the sourceforge FRS, I see no way that we can have package maintainers doing the file releases themselves. Regards David PS. Position of octave-forge release manager is available :-) -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
|
From: <so...@ha...> - 2007-05-22 09:21:36
|
David Bateman skrev: > I'd thought of that and have been working on the "make -C packages > compare" target of octave-forge. It now works correctly and the MD5 sums > of unmodified packages are respected, and the files to upload are placed > in the package directories. It also flags versioning issues (ie modified > packages without modifying the version number). Sounds good > I'd suggest a release of octave-forge in the next few days.. I'll be > using release-forge this time as the sourceforge FRS webpages are > horrible and the last release took me many hours to get right. We should > also do some package testing before the upload (at least run the "make > check" target). If others want to run "./configure; make all; make > check" on different platforms that would help.. This requires that all package dependencies of all packages are installed, right? I can only run the tests on Ubuntu on a x86, so I can't test any fancy hardware. But I'll try to run the tests tonight. > However, I would suggest keeping the supported version of Octave at > 2.9.10 as there has been little that has changed that should affect the > packaging itself.. Agreed. > In any case given the way the webpages are connected to the file > releases and the limitations of the sourceforge FRS, I see no way that > we can have package maintainers doing the file releases themselves. That kinda sucks! I guess the best we can do is then to announce (say) one week in advance that we release, and then package managers can fix anything that needs fixing (and possibly ask for more time)., > PS. Position of octave-forge release manager is available :-) I'd really hate to take on this position, but I guess everybody feels this way :-) If nobody else volunteers I guess I could be persuaded, but as I said, I really don't want to... Søren |
|
From: David B. <Dav...@mo...> - 2007-05-22 12:04:03
|
Søren Hauberg wrote: >> I'd suggest a release of octave-forge in the next few days.. I'll be >> using release-forge this time as the sourceforge FRS webpages are >> horrible and the last release took me many hours to get right. We should >> also do some package testing before the upload (at least run the "make >> check" target). If others want to run "./configure; make all; make >> check" on different platforms that would help.. > This requires that all package dependencies of all packages are > installed, right? I can only run the tests on Ubuntu on a x86, so I > can't test any fancy hardware. But I'll try to run the tests tonight. > The packages I have some sort of issue with can be checked with ./autogen.sh ./configure make all cd packages make checkpkg/extra/java-1.0.1.tar.gz make checkpkg/extra/secs1d-0.0.1.tar.gz make checkpkg/extra/secs2d-0.0.1.tar.gz make checkpkg/extra/windows-1.0.0.tar.gz make checkpkg/main/fixed-0.7.1.tar.gz make checkpkg/main/specfun-1.0.0.tar.gz >> In any case given the way the webpages are connected to the file >> releases and the limitations of the sourceforge FRS, I see no way that >> we can have package maintainers doing the file releases themselves. > That kinda sucks! I guess the best we can do is then to announce (say) > one week in advance that we release, and then package managers can fix > anything that needs fixing (and possibly ask for more time)., That is kind of what I thought as well. > >> PS. Position of octave-forge release manager is available :-) > I'd really hate to take on this position, but I guess everybody feels > this way :-) If nobody else volunteers I guess I could be persuaded, > but as I said, I really don't want to... Yeah, I have rather too many things on the go at the same time, and would love to overload this. Even release testing and fixing the errors given by "make check" would be great.. In any case I've tried to keep the release process up to date in the file admin/release.sh, though its now rather complex and involved a large number of manual steps.. Cheers David -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
|
From: <so...@ha...> - 2007-05-22 20:23:16
|
David Bateman skrev: > Søren Hauberg wrote: >>> I'd suggest a release of octave-forge in the next few days.. I'll be >>> using release-forge this time as the sourceforge FRS webpages are >>> horrible and the last release took me many hours to get right. We should >>> also do some package testing before the upload (at least run the "make >>> check" target). If others want to run "./configure; make all; make >>> check" on different platforms that would help.. >> This requires that all package dependencies of all packages are >> installed, right? I can only run the tests on Ubuntu on a x86, so I >> can't test any fancy hardware. But I'll try to run the tests tonight. >> > > The packages I have some sort of issue with can be checked with > > ./autogen.sh > ./configure > make all > cd packages > make checkpkg/extra/java-1.0.1.tar.gz > make checkpkg/extra/secs1d-0.0.1.tar.gz > make checkpkg/extra/secs2d-0.0.1.tar.gz > make checkpkg/extra/windows-1.0.0.tar.gz > make checkpkg/main/fixed-0.7.1.tar.gz > make checkpkg/main/specfun-1.0.0.tar.gz I've just started building octave-forge, but things break for me during "make all". I get a large amount of " *** extra/civil-engineering-1.0.0.tar.gz differs from uploaded version but has same version " errors (for many (all?) packages, not just the civil-engineering). It finally exits with " *** Please fix the version numbering noted above before proceding make: *** [real_compare] Fejl 255 " (Fejl = Error, it's Danish). It seems that newly created packages doesn't include a ChangeLog file, whereas the ones downloaded from SF does. Søren |
|
From: David B. <Dav...@mo...> - 2007-05-23 08:09:21
|
Søren Hauberg wrote: >> >> ./autogen.sh >> ./configure >> make all >> cd packages >> make checkpkg/extra/java-1.0.1.tar.gz >> make checkpkg/extra/secs1d-0.0.1.tar.gz >> make checkpkg/extra/secs2d-0.0.1.tar.gz >> make checkpkg/extra/windows-1.0.0.tar.gz >> make checkpkg/main/fixed-0.7.1.tar.gz >> make checkpkg/main/specfun-1.0.0.tar.gz I believe all of the errors for these are fixed in the CVS of octave-forge, though I can't easily test the windows of MacOS packages.. If someone would test those it would help.. I propose that in a couple of days to do an Octave-forge release.. Regards David -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
|
From: Thomas T. <Tho...@gm...> - 2007-05-23 16:10:49
|
David Bateman schrieb:
> Søren Hauberg wrote:
>>> ./autogen.sh
>>> ./configure
>>> make all
>>> cd packages
>>> make checkpkg/extra/java-1.0.1.tar.gz
>>> make checkpkg/extra/secs1d-0.0.1.tar.gz
>>> make checkpkg/extra/secs2d-0.0.1.tar.gz
>>> make checkpkg/extra/windows-1.0.0.tar.gz
>>> make checkpkg/main/fixed-0.7.1.tar.gz
>>> make checkpkg/main/specfun-1.0.0.tar.gz
> I believe all of the errors for these are fixed in the CVS of
> octave-forge, though I can't easily test the windows of MacOS packages..
> If someone would test those it would help.. I propose that in a couple
> of days to do an Octave-forge release..
>
> Regards
> David
MacOSX 10.4.9 IA32 (with macports dependencies), brief report:
./make distclean is ok.
./autogen.sh is ok.
./configure is ok.
./make all hangs up there and nothing happens anymore (can somebody please have
a look at the Makefile)
latex mintoolkit.tex > /dev/null 2>&1
PRESSING <CTRL><D> (I think I break the call of 'latex') continues package
creation resp. './make all' again!
The next errors then are
latex vrml-mini-howto-1.tex > /dev/null 2>&1
dvipdf vrml-mini-howto-1.dvi
latex2html -verbosity=0 -local_icons vrml-mini-howto-1.tex
make[4]: latex2html: Command not found
make[4]: *** [html/index.html] Error 127
make[4]: Target `all' not remade because of errors.
make[3]: *** [pre-pkg] Error 2
make[2]: *** [dopkg/vrml] Error 2
So we should change build process resp. the html creation of the vrml package.
Next I did
cd packages
find . -iname *tar.gz > files.sh
octave --version
GNU Octave, version 2.9.10
Copyright (C) 2007 John W. Eaton and others.
It is the real octave-2.9.10.tar.gz sources that I've used - no CVS updates. I
modified the files.sh file (see attachment)
sh ./files.sh 1>msg.txt 2>err.txt
I know, my own package fails - this is the problem of octave-2.9.10. David told
me he solved the pkg.m because of the mex-file creations. I attached msg.txt and
err.txt to this email. I maybe can check against octave-2.9.12 and have a look
for further details tomorrow. I'm sorry to be so shortspoken but I'm in big a
hurry, Thomas.
|
|
From: dbateman <dba...@fr...> - 2007-05-23 18:40:30
|
Ok so there are quite a few problems under MacOSX Thomas Treichl wrote: > > > ./make all hangs up there and nothing happens anymore (can somebody please > have > a look at the Makefile) > > latex mintoolkit.tex > /dev/null 2>&1 > I think I'll ignore this on as its only a problem for the person packaging the code and not for the package itself. I can correctly build the package under linux :-) > latex vrml-mini-howto-1.tex > /dev/null 2>&1 > dvipdf vrml-mini-howto-1.dvi > latex2html -verbosity=0 -local_icons vrml-mini-howto-1.tex > make[4]: latex2html: Command not found > make[4]: *** [html/index.html] Error 127 > make[4]: Target `all' not remade because of errors. > make[3]: *** [pre-pkg] Error 2 > make[2]: *** [dopkg/vrml] Error 2 > I imagine installing latex2html will fix this. Again I'll ignore this as its only an issue for the person packaging octave-forge not with the packages themselves. > cd packages > find . -iname *tar.gz > files.sh > octave --version > GNU Octave, version 2.9.10 > Copyright (C) 2007 John W. Eaton and others. > Using 2.9.12 would be better, as the .mex extension is accepted and there are a couple of other minor API changes. But this will be fine for the tests that I couldn't test with 2.9.12 > make checkpkg/extra/-.tar.gz > This is problem because of the packaging failure will "make all". I wouldn't worry too much about this. > make checkpkg/extra/mapping-1.0.0.tar.gz > make checkpkg/extra/nan-1.0.0.tar.gz > Where is the check for the macosx package? This is the one I really can't check otherwise... > Building main/combinatorics-1.0.0.tar.gz and its dependencies > Running test code for main/combinatorics-1.0.0.tar.gz and its dependencies > code/combinatorics-1.0.0/inst [tests 1 of 1 files] > ---> success > code/combinatorics-1.0.0/src [tests 1 of 1 files] > ---> passes 8 out of 10 tests > passes 9 out of 11 tests > There are failed tests above. Can you look in the fntests.log file and see what they are? > Building main/communications-1.0.1.tar.gz and its dependencies > Running test code for main/communications-1.0.1.tar.gz and its > dependencies > code/communications-1.0.1/inst [tests 17 of 65 files] > > << Random Signals Package >> > Signal Creation: PASSED > Signal Analysis: PASSED > > << Source Coding Package >> > PCM Functions: Not tested > Quantization Functions: PASSED > > << Block Coding Package >> > Cyclic Coding: PASSED > Hamming Coding: PASSED > BCH Coding: error: `glog' undefined near > line 38 column 7 > > ---> passes 58 out of 62 tests > There are failed tests here as well. However, it appears that the the problem here is bizarre. It appears that gf.oct either hasn't been built or the PKG_ADD file that autoload glog to gf hasn't been sourced. I suspect the later, as the build seems to have terminated correctly. > Building main/fixed-0.7.1.tar.gz and its dependencies > Running test code for main/fixed-0.7.1.tar.gz and its dependencies > code/fixed-0.7.1/examples [tests 0 of 4 files] > ---> success > code/fixed-0.7.1/inst [tests 1 of 6 files] > > << Fixed Point Load Type >> > error: Can not find fixed point type > This seems to be the same problem as above. Most strange.. I think I'll need a little bit more information to help debug this one.. > code/image-1.0.1/inst [tests 30 of 84 files] > ---> passes 222 out of 223 tests > I don't think this failure is expected. Chek fntests.log for more info.. > Building main/parallel-1.0.0.tar.gz and its dependencies > Failed to build main/parallel-1.0.0.tar.gz!!! > can you run cd packages make checkpkg/main/parallel-1.0.0.tar.gz and then look in sandbox/baddeps and try and see what didn't build. Alternatively just try and install the parallel pakage like octave:1> pkg install parallel-1.0.0.tar.gz it is unclear what is the issue here. > Building main/statistics-1.0.1.tar.gz and its dependencies > Running test code for main/statistics-1.0.1.tar.gz and its dependencies > code/statistics-1.0.1/inst [tests 28 of 55 files] > ---> passes 56 out of 58 tests > passes 56 out of 58 tests > If I remember correctly these errors aren't important, but again check fntests.log to see what they were. > > Building main/symbolic-1.0.1.tar.gz and its dependencies > Failed to build main/symbolic-1.0.1.tar.gz!!! > I image you don't have CLN and ginac installed. So this one doesn't bother me. > error: feval: the symbol `java_exit' is not valid as a function > error: feval: the symbol `java_exit' is not valid as a function > There is a problem with the java package.. I'm not going to chase this one. > sh: line 1: gmsh: command not found > sh: line 1: dx: command not found > You are missing some external dependencies for secs2d, this is also not an issue. > warning: autoload: `' is not an absolute file name > These are other warnings like it are just a consequence of the addpath on sandbox/code and can be ignored.. There is probably a way of getting rid of this if I could be bothered... In any case, they have no consequence on the testing. > panic: Bus error -- stopping myself... > /Users/Thomas/Developement/forge/admin/run_forge: line 62: 1300 Bus error > $* > Huhh.. This I don't see and have no idea why you do.. > warning: grpdelay: setting group delay to 0 at singularity > warning: grpdelay: setting group delay to 0 at singularity > warning: grpdelay: setting group delay to 0 at singularity > warning: grpdelay: setting group delay to 0 at singularity > warning: XXX FIXME XXX Tsq return from princomp fails some tests > warning: XXX FIXME XXX Tsq return from princomp fails some tests > warning: inverse: matrix singular to machine precision, rcond = 0 > Known and documented issue in the signal toolbox, someone else can fix it if they like :-) Cheers David -- View this message in context: http://www.nabble.com/New-release--tf3794959.html#a10770225 Sent from the octave-dev mailing list archive at Nabble.com. |
|
From: Thomas T. <Tho...@gm...> - 2007-05-24 18:41:06
|
I switched to octave-2.9.12 - let's have a look what is going wrong from the
things that I've already pointed out. I started from the very beginning...
Development/forge Thomas$ octave --version
GNU Octave, version 2.9.12+
Development/forge Thomas$ make distclean
Development/forge Thomas$ ./autogen.sh
Development/forge Thomas$ ./configure && make all
>> I think I'll ignore this on as its only a problem for the person packaging
>> the code and not for the package itself. I can correctly build the package
>> under linux :-)
>
> I won't ignore it and will have a look for it the following days even if I'm not
> the person who packages this ;) It should be fixed - fixing bugs is a good
> thing, don't you think?
Fixing that one would be easy:
(/opt/local/share/texmf-dist/tex/latex/graphics/dvips.def)))
! LaTeX Error: File `html.sty' not found.
Type X to quit or <RETURN> to proceed,
or enter new name. (Default extension: sty)
copying a html.sty file from the internet to the local directory
'main/optim/doc' would fix this problem. It seems that the standard latex
package (or dvips or whatever) doesn't include the html package.
>>> latex vrml-mini-howto-1.tex > /dev/null 2>&1
>>> dvipdf vrml-mini-howto-1.dvi
>>> latex2html -verbosity=0 -local_icons vrml-mini-howto-1.tex
>>> make[4]: latex2html: Command not found
>>> make[4]: *** [html/index.html] Error 127
>>> make[4]: Target `all' not remade because of errors.
>>> make[3]: *** [pre-pkg] Error 2
>>> make[2]: *** [dopkg/vrml] Error 2
>>
>> I imagine installing latex2html will fix this. Again I'll ignore this as its
>> only an issue for the person packaging octave-forge not with the packages
>> themselves.
>
> Maybe we can do something else instead of hard-coding latex2html in the Makefile
> - as you do in the comm-package and I do in the odepkg?! What if a user (not
> just Mac, Linux also or maybe even Win) should ever do these tests once again
> and he fails because of latex2html - I mean shouldn't we have bigger problems
> like that?
Meanwhile I fixed this and installed latex2html because of the OSX package that
fails if latex2html is missing...
>>> make checkpkg/extra/mapping-1.0.0.tar.gz
>>> make checkpkg/extra/nan-1.0.0.tar.gz
>> Where is the check for the macosx package? This is the one I really can't
>> check otherwise...
>
> Good question that I actually can't answer. There is no macosx package in my
> packages/extra directory. Ok, I go searching for it...
extra/MacOSX Thomas$ make
latex2html -verbosity=0 -local_icons ofosx.tex
make[1]: latex2html: Command not found
make[1]: *** [html/index.html] Error 127
make: *** [pre-pkg] Error 2
extra/MacOSX Thomas$
After installing latex2html I could find it...
forge/packages Thomas$ make checkpkg/extra/macosx-1.0.1.tar.gz
Building extra/macosx-1.0.1.tar.gz and its dependencies
Running test code for extra/macosx-1.0.1.tar.gz and its dependencies
code/macosx-1.0.1/src [tests 0 of 1 files]
---> success
passes 0 out of 0 tests
>>> Building main/combinatorics-1.0.0.tar.gz and its dependencies
>>> Running test code for main/combinatorics-1.0.0.tar.gz and its dependencies
>>> code/combinatorics-1.0.0/inst [tests 1 of 1 files]
>>> ---> success
>>> code/combinatorics-1.0.0/src [tests 1 of 1 files]
>>> ---> passes 8 out of 10 tests
>>> passes 9 out of 11 tests
>> There are failed tests above. Can you look in the fntests.log file and see
>> what they are?
>
> Voila,
>
> >>>>> processing
> /Users/Thomas/Development/forge/packages/sandbox/code/combinatorics-1.0.0/inst/combs.m
> >>>>> processing
> /Users/Thomas/Development/forge/packages/sandbox/code/combinatorics-1.0.0/src/partint.cc
> ***** assert(partcnt(1), 1);
> !!!!! test failed
> error: `partcnt' undefined near line 14 column 8
> ***** assert(partcnt(17), 297);
> !!!!! test failed
> error: `partcnt' undefined near line 14 column 8
Seems to be a octave-2.9.10 problem, see the output for 2.9.12:
Running test code for main/combinatorics-1.0.0.tar.gz and its dependencies
code/combinatorics-1.0.0/inst [tests 1 of 1 files]
---> success
code/combinatorics-1.0.0/src [tests 1 of 1 files]
---> success
passes 11 out of 11 tests
>> Building main/communications-1.0.1.tar.gz and its dependencies
>>> Running test code for main/communications-1.0.1.tar.gz and its
>>> dependencies
>>> code/communications-1.0.1/inst [tests 17 of 65 files]
>>>
>>> << Random Signals Package >>
>>> Signal Creation: PASSED
>>> Signal Analysis: PASSED
>>>
>>> << Source Coding Package >>
>>> PCM Functions: Not tested
>>> Quantization Functions: PASSED
>>>
>>> << Block Coding Package >>
>>> Cyclic Coding: PASSED
>>> Hamming Coding: PASSED
>>> BCH Coding: error: `glog' undefined near
>>> line 38 column 7
>>>
>>> ---> passes 58 out of 62 tests
>> There are failed tests here as well. However, it appears that the the
>> problem here is bizarre. It appears that gf.oct either hasn't been built or
>> the PKG_ADD file that autoload glog to gf hasn't been sourced. I suspect the
>> later, as the build seems to have terminated correctly.
>
> Yes the build has terminated correctly, here is the output of the fntest.log:
>
> >>>>> processing
> /Users/Thomas/Development/forge/packages/sandbox/code/communications-1.0.1/inst/pamdemod.m
> ***** assert (pamdemod([-7:2:7],8,0,'Bin'),[0:7])
> !!!!! test failed
>
> Invalid call to genqamdemod. Correct usage is:
>
> -- Loadable Function: Y = genqamdemod (X, C)
>
>
> Additional help for built-in functions and operators is
> available in the on-line version of the manual. Use the command
> `doc <topic>' to search the manual index.
>
> Help and information about Octave is also available on the WWW
> at http://www.octave.org and via the he...@oc...
> mailing list.
>
> ***** assert (pamdemod([-7:2:7],8,0,'Gray'),[0 1 3 2 6 7 5 4])
> !!!!! test failed
>
> Invalid call to genqamdemod. Correct usage is:
>
> -- Loadable Function: Y = genqamdemod (X, C)
>
>
> Additional help for built-in functions and operators is
> available in the on-line version of the manual. Use the command
> `doc <topic>' to search the manual index.
>
> Help and information about Octave is also available on the WWW
> at http://www.octave.org and via the he...@oc...
> mailing list.
Better with 2.9.12:
<< Block Coding Package >>
Cyclic Coding: PASSED
Hamming Coding: PASSED
BCH Coding: PASSED
Reed-Solomon Coding: PASSED
> >>>>> processing
> /Users/Thomas/Development/forge/packages/sandbox/code/communications-1.0.1/inst/pammod.m
> ***** assert(round(pammod([0:7],8,0,'Bin')),[-7:2:7])
> !!!!! test failed
Solved and fixed with 2.9.12. No fails anymore.
>>> Building main/fixed-0.7.1.tar.gz and its dependencies
>>> Running test code for main/fixed-0.7.1.tar.gz and its dependencies
>>> code/fixed-0.7.1/examples [tests 0 of 4 files]
>>> ---> success
>>> code/fixed-0.7.1/inst [tests 1 of 6 files]
>>>
>>> << Fixed Point Load Type >>
>>> error: Can not find fixed point type
> >
>> This seems to be the same problem as above. Most strange.. I think I'll need
>> a little bit more information to help debug this one..
>
> Hhmmm... nothing strange in the fntest.log file. You have some tests for me that
> have nothing to do with packaging, or do I have to go through each file manually?
Looks good now with 2.9.12:
code/fixed-0.7.1/examples [tests 0 of 4 files]
---> success
code/fixed-0.7.1/inst [tests 1 of 6 files]
<< Fixed Point Load Type >>
Found Fixed Point Toolbox (version 0.7.1)
<< Fixed Point Creation >>
Scalar Creation: PASSED
Matrix Creation: PASSED
Complex Scalar Creation: PASSED
Complex Matrix Creation: PASSED
Indexed Access of Scalars: PASSED
Indexed Access of Matrices: PASSED
<< Fixed Point Operators >>
Logical Operators: PASSED
Unary Operators: PASSED
Arithmetic Operators: PASSED
<< Fixed Point Functions >>
Rounding Functions: PASSED
Matrix Sum and Product Functions: PASSED
Miscellaneous Functions: PASSED
Exponential Functions: PASSED
Trigonometric Functions: PASSED
---> success
code/fixed-0.7.1/src [tests 0 of 35 files]
---> success
code/fixed-0.7.1/src/int [tests 0 of 1 files]
---> success
passes 1 out of 1 tests
>>> code/image-1.0.1/inst [tests 30 of 84 files]
>>> ---> passes 222 out of 223 tests
>>>
>> I don't think this failure is expected. Chek fntests.log for more info..
>
> >>>>> processing
> /Users/Thomas/Developement/forge/packages/sandbox/code/image-1.0.1/inst/imrotate.m
> ***** test
> ## Verify minimal loss across six rotations that add up to 360 +/- 1 deg.:
> methods = { "nearest", "bilinear", "bicubic", "Fourier" };
> angles = [ 59 60 61 ];
> tolerances = [ 7.4 8.5 8.6 # nearest
> 3.5 3.1 3.5 # bilinear
> 2.7 2.0 2.7 # bicubic
> 2.7 1.6 2.8 ]/8; # Fourier
>
> # This is peaks(50) without the dependency on the plot package
> x = y = linspace(-3,3,50);
> [X,Y] = meshgrid(x,y);
> x = 3*(1-X).^2.*exp(-X.^2 - (Y+1).^2) \
> - 10*(X/5 - X.^3 - Y.^5).*exp(-X.^2-Y.^2) \
> - 1/3*exp(-(X+1).^2 - Y.^2);
>
> x -= min(x(:)); # Fourier does not handle neg. values well
> x = x./max(x(:));
> for m = 1:(length(methods))
> y = x;
> for i = 1:5
> y = imrotate(y, 60, methods{m}, "crop", 0);
> end
> for a = 1:(length(angles))
> assert(norm((x - imrotate(y, angles(a), methods{m}, "crop", 0))
> (10:40, 10:40)) < tolerances(m,a));
> end
> end
> !!!!! test failed
> error: `im' undefined near line 42 column 35
This fail is still there and did not disappear with 2.9.12:
>>> Building main/parallel-1.0.0.tar.gz and its dependencies
>>> Failed to build main/parallel-1.0.0.tar.gz!!!
>>>
>> can you run
>>
>> cd packages
>> make checkpkg/main/parallel-1.0.0.tar.gz
>>
>> and then look in sandbox/baddeps and try and see what didn't build.
>> Alternatively just try and install the parallel pakage like
>
> Thomas$ cat sandbox/baddeps
> mkoctfile: stripping disabled on this platform
> mkoctfile: stripping disabled on this platform
> connect.cc: In function 'octave_value_list Fconnect(const octave_value_list&, int)':
> connect.cc:222: error: '__BYTE_ORDER' was not declared in this scope
> make[1]: *** [connect.oct] Error 1
> error: 'make' returned the following error: mkoctfile -s sclose.cc
> mkoctfile -s connect.cc
I can have a further look at this tomorrow (somebody has ideas?)
forge/main/parallel/src Thomas$ make
mkoctfile -s sclose.cc
mkoctfile: stripping disabled on this platform
mkoctfile -s connect.cc
mkoctfile: stripping disabled on this platform
connect.cc: In function 'octave_value_list Fconnect(const octave_value_list&, int)':
connect.cc:222: error: '__BYTE_ORDER' was not declared in this scope
make: *** [connect.oct] Error 1
>>> Building main/statistics-1.0.1.tar.gz and its dependencies
>>> Running test code for main/statistics-1.0.1.tar.gz and its dependencies
>>> code/statistics-1.0.1/inst [tests 28 of 55 files]
>>> ---> passes 56 out of 58 tests
>>> passes 56 out of 58 tests
> >>>>> processing
> /Users/Thomas/Developement/forge/packages/sandbox/code/statistics-1.0.1/inst/princomp.m
> ***** assert(Tsq,1,10*eps);
> !!!!! test failed
> error: assert (Tsq,1,10 * eps) expected
> 1
> but got
> NaN
> NaN
> Dimensions don't match
> shared variables {
> pc =
>
> -0.70711 0.70711 0.00000
> 0.70711 0.70711 0.00000
> -0.00000 0.00000 1.00000
>
> z =
>
> 0.70711 0.00000 0.00000
> -0.70711 -0.00000 0.00000
>
> w =
>
> 1
> 0
> 0
>
> Tsq =
>
> NaN
> NaN
>
> m =
>
> -0.70711 0.70711 0.00000
> 0.70711 0.70711 0.00000
> -0.00000 0.00000 1.00000
>
> x =
>
> 1 2 3
> 2 1 3
>
> }
>
> >>>>> processing
> /Users/Thomas/Development/forge/packages/sandbox/code/statistics-1.0.1/inst/regress.m
> ***** test
> % Longley data from the NIST Statistical Reference Dataset
> Z = [ 60323 83.0 234289 2356 1590 107608 1947
> 61122 88.5 259426 2325 1456 108632 1948
> 60171 88.2 258054 3682 1616 109773 1949
> 61187 89.5 284599 3351 1650 110929 1950
> 63221 96.2 328975 2099 3099 112075 1951
> 63639 98.1 346999 1932 3594 113270 1952
> 64989 99.0 365385 1870 3547 115094 1953
> 63761 100.0 363112 3578 3350 116219 1954
> 66019 101.2 397469 2904 3048 117388 1955
> 67857 104.6 419180 2822 2857 118734 1956
> 68169 108.4 442769 2936 2798 120445 1957
> 66513 110.8 444546 4681 2637 121950 1958
> 68655 112.6 482704 3813 2552 123366 1959
> 69564 114.2 502601 3931 2514 125368 1960
> 69331 115.7 518173 4806 2572 127852 1961
> 70551 116.9 554894 4007 2827 130081 1962 ];
> % Results certified by NIST using 500 digit arithmetic
> % b and standard error in b
> V = [ -3482258.63459582 890420.383607373
> 15.0618722713733 84.9149257747669
> -0.358191792925910E-01 0.334910077722432E-01
> -2.02022980381683 0.488399681651699
> -1.03322686717359 0.214274163161675
> -0.511041056535807E-01 0.226073200069370
> 1829.15146461355 455.478499142212 ];
> Rsq = 0.995479004577296;
> F = 330.285339234588;
> y = Z(:,1); X = [ones(rows(Z),1), Z(:,2:end)];
> alpha = 0.05;
> [b, bint, r, rint, stats] = regress (y, X, alpha);
> assert(b,V(:,1),3e-6);
> assert(stats(1),Rsq,3e-13);
> assert(stats(2),F,3e-8);
> assert(((bint(:,1)-bint(:,2))/2)/t_inv(alpha/2,9),V(:,2),-1.e-5);
> !!!!! test failed
> error: assert (stats (1),Rsq,3e-13) expected
> 0.99548
> but got
> 0.99548
> maximum absolute error 3.42504e-13 exceeds tolerance 3e-13
>> If I remember correctly these errors aren't important, but again check
>> fntests.log to see what they were.
This one is still there in 2.9.12.
>> error: feval: the symbol `java_exit' is not valid as a function
>> error: feval: the symbol `java_exit' is not valid as a function
>>
> There is a problem with the java package.. I'm not going to chase this one.
Fixed with 2.9.12 as Michael already told us on the list ;)
Running test code for extra/java-1.0.1.tar.gz and its dependencies
code/java-1.0.1/inst [tests 0 of 3 files]
---> success
code/java-1.0.1/src [tests 0 of 1 files]
---> success
passes 0 out of 0 tests
>>> sh: line 1: gmsh: command not found
>>> sh: line 1: dx: command not found
>> You are missing some external dependencies for secs2d, this is also not an
>> issue.
>
> Yes this might be...
>
>>> warning: autoload: `' is not an absolute file name
>> These are other warnings like it are just a consequence of the addpath on
>> sandbox/code and can be ignored.. There is probably a way of getting rid of
>> this if I could be bothered... In any case, they have no consequence on the
>> testing.
>>
>>> panic: Bus error -- stopping myself...
>>> /Users/Thomas/Development/forge/admin/run_forge: line 62: 1300 Bus error
>>> $*
>> Huhh.. This I don't see and have no idea why you do..
>
> Yupp, odepkg - as I've expected with 2.9.10 ;)
Running test code for main/odepkg-0.3.0.tar.gz and its dependencies
code/odepkg-0.3.0/inst [tests 21 of 33 files]
---> success
passes 212 out of 212 tests
>>> warning: grpdelay: setting group delay to 0 at singularity
>>> warning: grpdelay: setting group delay to 0 at singularity
>>> warning: grpdelay: setting group delay to 0 at singularity
>>> warning: grpdelay: setting group delay to 0 at singularity
>>> warning: XXX FIXME XXX Tsq return from princomp fails some tests
>>> warning: XXX FIXME XXX Tsq return from princomp fails some tests
>>> warning: inverse: matrix singular to machine precision, rcond = 0
>> Known and documented issue in the signal toolbox, someone else can fix it if
>> they like :-)
>>
>> Cheers
>> David
>
> So, next step is waiting until 2.9.12 has finished compilation and then I reply
> once again. Thomas
So looks already better, but some fails are left.
Thomas
|
|
From: David B. <ad...@gm...> - 2007-05-24 19:36:37
|
Thomas Treichl wrote:
<Lots of fixed bugs removed that I'm very happy about :-)>
>
>>>> code/image-1.0.1/inst [tests 30 of 84 files]
>>>> ---> passes 222 out of 223
>>>> tests
>>>>
>>> I don't think this failure is expected. Chek fntests.log for more
>>> info..
>>
>> >>>>> processing
>> /Users/Thomas/Developement/forge/packages/sandbox/code/image-1.0.1/inst/imrotate.m
>>
>> ***** test
>> ## Verify minimal loss across six rotations that add up to 360 +/-
>> 1 deg.:
>> methods = { "nearest", "bilinear", "bicubic", "Fourier" };
>> angles = [ 59 60 61 ];
>> tolerances = [ 7.4 8.5 8.6 # nearest
>> 3.5 3.1 3.5 # bilinear
>> 2.7 2.0 2.7 # bicubic
>> 2.7 1.6 2.8 ]/8; # Fourier
>>
>> # This is peaks(50) without the dependency on the plot package
>> x = y = linspace(-3,3,50);
>> [X,Y] = meshgrid(x,y);
>> x = 3*(1-X).^2.*exp(-X.^2 - (Y+1).^2) \
>> - 10*(X/5 - X.^3 - Y.^5).*exp(-X.^2-Y.^2) \
>> - 1/3*exp(-(X+1).^2 - Y.^2);
>>
>> x -= min(x(:)); # Fourier does not handle neg. values well
>> x = x./max(x(:));
>> for m = 1:(length(methods))
>> y = x;
>> for i = 1:5
>> y = imrotate(y, 60, methods{m}, "crop", 0);
>> end
>> for a = 1:(length(angles))
>> assert(norm((x - imrotate(y, angles(a), methods{m}, "crop", 0))
>> (10:40, 10:40)) < tolerances(m,a));
>> end
>> end
>> !!!!! test failed
>> error: `im' undefined near line 42 column 35
>
> This fail is still there and did not disappear with 2.9.12:
>
The is a bug in imrotate_Fourier.m. I fixed this in the CVS..
>
>>>> Building main/parallel-1.0.0.tar.gz and its dependencies
>>>> Failed to build main/parallel-1.0.0.tar.gz!!!
>>>>
>>> can you run
>>>
>>> cd packages
>>> make checkpkg/main/parallel-1.0.0.tar.gz
>>>
>>> and then look in sandbox/baddeps and try and see what didn't build.
>>> Alternatively just try and install the parallel pakage like
>>
>> Thomas$ cat sandbox/baddeps
>> mkoctfile: stripping disabled on this platform
>> mkoctfile: stripping disabled on this platform
>> connect.cc: In function 'octave_value_list Fconnect(const
>> octave_value_list&, int)':
>> connect.cc:222: error: '__BYTE_ORDER' was not declared in this scope
>> make[1]: *** [connect.oct] Error 1
>> error: 'make' returned the following error: mkoctfile -s sclose.cc
>> mkoctfile -s connect.cc
>
> I can have a further look at this tomorrow (somebody has ideas?)
> forge/main/parallel/src Thomas$ make
> mkoctfile -s sclose.cc
> mkoctfile: stripping disabled on this platform
> mkoctfile -s connect.cc
> mkoctfile: stripping disabled on this platform
> connect.cc: In function 'octave_value_list Fconnect(const
> octave_value_list&, int)':
> connect.cc:222: error: '__BYTE_ORDER' was not declared in this scope
> make: *** [connect.oct] Error 1
Can you try adding
#include <endian.h>
to the top of connect.cc and rebuild
>
>
>>>> Building main/statistics-1.0.1.tar.gz and its dependencies
>>>> Running test code for main/statistics-1.0.1.tar.gz and its
>>>> dependencies
>>>> code/statistics-1.0.1/inst [tests 28 of 55 files]
>>>> ---> passes 56 out of 58
>>>> tests
>>>> passes 56 out of 58 tests
>> >>>>> processing
>> /Users/Thomas/Developement/forge/packages/sandbox/code/statistics-1.0.1/inst/princomp.m
>>
>> ***** assert(Tsq,1,10*eps);
>> !!!!! test failed
>> error: assert (Tsq,1,10 * eps) expected
>> 1
>> but got
>> NaN
>> NaN
>> Dimensions don't match
>> shared variables {
>> pc =
>>
>> -0.70711 0.70711 0.00000
>> 0.70711 0.70711 0.00000
>> -0.00000 0.00000 1.00000
>>
>> z =
>>
>> 0.70711 0.00000 0.00000
>> -0.70711 -0.00000 0.00000
>>
>> w =
>>
>> 1
>> 0
>> 0
>>
>> Tsq =
>>
>> NaN
>> NaN
>>
>> m =
>>
>> -0.70711 0.70711 0.00000
>> 0.70711 0.70711 0.00000
>> -0.00000 0.00000 1.00000
>>
>> x =
>>
>> 1 2 3
>> 2 1 3
>>
>> }
>>
>> >>>>> processing
>> /Users/Thomas/Development/forge/packages/sandbox/code/statistics-1.0.1/inst/regress.m
>>
>> ***** test
>> % Longley data from the NIST Statistical Reference Dataset
>> Z = [ 60323 83.0 234289 2356 1590 107608 1947
>> 61122 88.5 259426 2325 1456 108632 1948
>> 60171 88.2 258054 3682 1616 109773 1949
>> 61187 89.5 284599 3351 1650 110929 1950
>> 63221 96.2 328975 2099 3099 112075 1951
>> 63639 98.1 346999 1932 3594 113270 1952
>> 64989 99.0 365385 1870 3547 115094 1953
>> 63761 100.0 363112 3578 3350 116219 1954
>> 66019 101.2 397469 2904 3048 117388 1955
>> 67857 104.6 419180 2822 2857 118734 1956
>> 68169 108.4 442769 2936 2798 120445 1957
>> 66513 110.8 444546 4681 2637 121950 1958
>> 68655 112.6 482704 3813 2552 123366 1959
>> 69564 114.2 502601 3931 2514 125368 1960
>> 69331 115.7 518173 4806 2572 127852 1961
>> 70551 116.9 554894 4007 2827 130081 1962 ];
>> % Results certified by NIST using 500 digit arithmetic
>> % b and standard error in b
>> V = [ -3482258.63459582 890420.383607373
>> 15.0618722713733 84.9149257747669
>> -0.358191792925910E-01 0.334910077722432E-01
>> -2.02022980381683 0.488399681651699
>> -1.03322686717359 0.214274163161675
>> -0.511041056535807E-01 0.226073200069370
>> 1829.15146461355 455.478499142212 ];
>> Rsq = 0.995479004577296;
>> F = 330.285339234588;
>> y = Z(:,1); X = [ones(rows(Z),1), Z(:,2:end)];
>> alpha = 0.05;
>> [b, bint, r, rint, stats] = regress (y, X, alpha);
>> assert(b,V(:,1),3e-6);
>> assert(stats(1),Rsq,3e-13);
>> assert(stats(2),F,3e-8);
>> assert(((bint(:,1)-bint(:,2))/2)/t_inv(alpha/2,9),V(:,2),-1.e-5);
>> !!!!! test failed
>> error: assert (stats (1),Rsq,3e-13) expected
>> 0.99548
>> but got
>> 0.99548
>> maximum absolute error 3.42504e-13 exceeds tolerance 3e-13
> >> If I remember correctly these errors aren't important, but again check
> >> fntests.log to see what they were.
>
> This one is still there in 2.9.12.
Its a statistical test, so give how close it is to passing, I'm inclined
just to widen the error margin slightly.. And have done so..
Regards
D.
|
|
From: David B. <ad...@gm...> - 2007-05-24 19:48:30
Attachments:
patch.connect
|
David Bateman wrote: >> I can have a further look at this tomorrow (somebody has ideas?) >> forge/main/parallel/src Thomas$ make >> mkoctfile -s sclose.cc >> mkoctfile: stripping disabled on this platform >> mkoctfile -s connect.cc >> mkoctfile: stripping disabled on this platform >> connect.cc: In function 'octave_value_list Fconnect(const >> octave_value_list&, int)': >> connect.cc:222: error: '__BYTE_ORDER' was not declared in this scope >> make: *** [connect.oct] Error 1 >> > Can you try adding > > #include <endian.h> > > to the top of connect.cc and rebuild > > Could you also try the attached patch D. |
|
From: Carlo de F. <car...@gm...> - 2007-05-22 17:14:02
|
David,
On 22/mag/07, at 15:26, David Bateman wrote:
> Carlo de Falco wrote:
>> On 5/22/07, David Bateman <Dav...@mo...> wrote:
>>
>>> The packages I have some sort of issue with can be checked with
>>>
>>> ./autogen.sh
>>> ./configure
>>> make all
>>> cd packages
>>>
>>
>>
>>> make checkpkg/extra/secs1d-0.0.1.tar.gz
>>> make checkpkg/extra/secs2d-0.0.1.tar.gz
>>>
>>
>> This two packages are mine I'll have a look later today
>> to see what are the issues.
>>
> Maybe there are dependency issues on my installation. Haven't checked
> yet. I just flagged the packages that failed "make check" for me..
I ran make checkpkg/extra/secs2d-0.0.1.tar.gz checkpkg/extra/
secs1d-0.0.1.tar.gz
below is the output for secs2d, of course there is no failing test as
there is no test yet ;)
OTH there are warnings from addpath not finding some directories.
This is due to commands in PKG_ADD that add subdirectories of the
package directory to the path.
This works fine after installation but fails during make check
because the subdirectories are still inside
the directory inst.
Is there something I can do to get rid of suche warnings? Should I
just live with it?
Finally I don't understand why the commands in PKG_ADD are run so
many times...
Thanks,
c.
make checkpkg/extra/secs2d-0.0.1.tar.gz
Building extra/secs2d-0.0.1.tar.gz and its dependencies
gmsh version found:
dx version found:
Open Visualization Data Explorer Executive, version 04.4.4000
(13:45:09, May 13 2007)
Open Visualization Data Explorer User Interface, version 04.04.0004
(13:46:31, May 13 2007)
Open Visualization Data Explorer Script, 04.4.4000
gmsh version found:
dx version found:
Open Visualization Data Explorer Executive, version 04.4.4000
(13:45:09, May 13 2007)
Open Visualization Data Explorer User Interface, version 04.04.0004
(13:46:31, May 13 2007)
Open Visualization Data Explorer Script, 04.4.4000
Running test code for extra/secs2d-0.0.1.tar.gz and its dependencies
gmsh version found:
2.0.7
dx version found:
Open Visualization Data Explorer Executive, version 04.4.4000
(13:45:09, May 13 2007)
Open Visualization Data Explorer User Interface, version 04.04.0004
(13:46:31, May 13 2007)
Open Visualization Data Explorer Script, 04.4.4000
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/Utilities: No such file or directory
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/DDGOX: No such file or directory
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/ThDDGOX: No such file or directory
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/QDDGOX: No such file or directory
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/METLINES: No such file or directory
warning: addpath: /Users/carlo/Desktop/octave-forge/packages/sandbox/
code/secs2d-0.0.1/DDGOXT: No such file or directory
gmsh version found:
2.0.7
dx version found:
Open Visualization Data Explorer Executive, version 04.4.4000
(13:45:09, May 13 2007)
Open Visualization Data Explorer User Interface, version 04.04.0004
(13:46:31, May 13 2007)
Open Visualization Data Explorer Script, 04.4.4000
code/secs2d-0.0.1/inst/DDGOX [tests 0 of 6 files]
---> success
code/secs2d-0.0.1/inst/DDGOXT [tests 0 of 3 files]
---> success
code/secs2d-0.0.1/inst/METLINES [tests 0 of 2 files]
---> success
code/secs2d-0.0.1/inst/QDDGOX [tests 0 of 4 files]
---> success
code/secs2d-0.0.1/inst/ThDDGOX [tests 0 of 7 files]
---> success
code/secs2d-0.0.1/inst/Utilities [tests 0 of 42 files]
---> success
code/secs2d-0.0.1/src/Utilities [tests 0 of 4 files]
---> success
passes 0 out of 0 tests
|
|
From: dbateman <dba...@fr...> - 2007-05-22 19:53:31
|
Carlo de Falco-2 wrote: > > > I ran make checkpkg/extra/secs2d-0.0.1.tar.gz checkpkg/extra/ > secs1d-0.0.1.tar.gz > below is the output for secs2d, of course there is no failing test as > there is no test yet ;) > Yes, except that the src/Makefile should use "make" and not "Make". I presume you are using a platform where path and file names are case insensitive. I committed a patch > OTH there are warnings from addpath not finding some directories. > This is due to commands in PKG_ADD that add subdirectories of the > package directory to the path. > This works fine after installation but fails during make check > because the subdirectories are still inside > the directory inst. > Is there something I can do to get rid of suche warnings? Should I > just live with it? > Finally I don't understand why the commands in PKG_ADD are run so > many times... > This however I'm still not sure how to treat it. The issue is that some packages include additional test code that is not installed with the package, and so the "make check" code addpath the directories of the uninstalled package to have access to this additional test code. The easiest way to treat this would be in the PKG_ADD and PKG_DEL files and either suppress the harmless warning, or include a test that if we are in the uninstalled tree we don't addpath these additional directories. Given the arbitrary way in which the packager might include additional test code, I'm not sure I see another way of doing it in fact. D. -- View this message in context: http://www.nabble.com/New-release--tf3794959.html#a10751282 Sent from the octave-dev mailing list archive at Nabble.com. |
|
From: <so...@ha...> - 2007-05-23 19:10:02
|
dbateman skrev: > > > Søren Hauberg wrote: >> >> I've just started building octave-forge, but things break for me during >> "make all". I get a large amount of >> >> " >> *** extra/civil-engineering-1.0.0.tar.gz differs from uploaded version >> but has same version >> " >> >> errors (for many (all?) packages, not just the civil-engineering). It >> finally exits with >> >> >> " >> *** Please fix the version numbering noted above before proceding >> make: *** [real_compare] Fejl 255 >> " >> >> (Fejl = Error, it's Danish). It seems that newly created packages >> doesn't include a ChangeLog file, whereas the ones downloaded from SF >> does. >> >> Søren >> >> > > Are you perhaps trying to do the build on a machine without CVS access to > Sourceforge? You must have CVS access to sourceforge from the source tree > for "make all" to work. Okay, so I have my SSH key in place, and I can run cvs2cl without issues. I do however still get the same errors. If I run diff packages/compare/extra/civil-engineering-1.0.0.tar.gz packages/extra/civil-engineering-1.0.0.tar.gz I am told that the files differ. If I test the actual contents of the files they don't seem to differ. Is it possible that tar works different on your system than on mine? Perhaps different default compression levels? Søren |
|
From: <so...@ha...> - 2007-05-23 19:49:52
|
dbateman skrev: > Maybe, but not only. The main issue is that the packaging creates a > temporary directory in which the parts of the files to be packaged are > placed. The time stamps on these directories differ for the uploaded > versions and therefore the tarballs differ even if the contents don't. I see, that makes sense. > This was the major issue addressed in the "make -C packages compare" target > which should work and correctly identify that the packages don't differ... I tried running "make -C packages compare" and I get the same error (I'm cut'n'pasting the entire output to the end of this message). I guess I'm doing something wrong here, but I really don't know what. Anyway, perhaps it would be best to ignore my problems for now, as I guess you're more than busy right now :-) Søren sh@sh-laptop:~/Dokumenter/octave-forge/octave-forge$ make -C packages compare make: Går til katalog '/home/sh/Dokumenter/octave-forge/octave-forge/packages' --21:38:27-- http://octave.sourceforge.net/packages.md5 => `packages.md5' Løser octave.sourceforge.net... 66.35.250.209 Connecting to octave.sourceforge.net|66.35.250.209|:80... forbundet. HTTP forespørgsel sendt, afventer svar... 200 OK Længde: 6.482 (6.3K) [text/plain] 100%[====================================>] 6.482 33.68K/s 21:38:28 (33.66 KB/s) - `packages.md5' saved [6482/6482] --21:38:28-- http://downloads.sourceforge.net/octave/octave-forge-bundle-20070328.tar.gz?download => `octave-forge-bundle-20070328.tar.gz?download' Løser downloads.sourceforge.net... 66.35.250.203 Connecting to downloads.sourceforge.net|66.35.250.203|:80... forbundet. HTTP forespørgsel sendt, afventer svar... 302 Found Sted: http://dfn.dl.sourceforge.net/sourceforge/octave/octave-forge-bundle-20070328.tar.gz [omdirigeret] --21:38:28-- http://dfn.dl.sourceforge.net/sourceforge/octave/octave-forge-bundle-20070328.tar.gz => `octave-forge-bundle-20070328.tar.gz' Løser dfn.dl.sourceforge.net... 194.95.236.6 Connecting to dfn.dl.sourceforge.net|194.95.236.6|:80... forbundet. HTTP forespørgsel sendt, afventer svar... 200 OK Længde: 4.839.197 (4.6M) [application/x-gzip] 100%[====================================>] 4.839.197 238.45K/s ETA 00:00 21:38:48 (247.46 KB/s) - `octave-forge-bundle-20070328.tar.gz' saved [4839197/4839197] extra/macosx-1.0.1.tar.gz differs from uploaded version extra/symband-1.0.1.tar.gz differs from uploaded version *** extra/engine-1.0.0.tar.gz differs from uploaded version but has same version *** extra/graceplot-1.0.0.tar.gz differs from uploaded version but has same version *** extra/integration-1.0.0.tar.gz differs from uploaded version but has same version extra/java-1.0.1.tar.gz differs from uploaded version *** extra/pdb-1.0.0.tar.gz differs from uploaded version but has same version *** extra/tsa-3.10.0.tar.gz differs from uploaded version but has same version *** extra/windows-1.0.0.tar.gz differs from uploaded version but has same version extra/xraylib-1.0.1.tar.gz differs from uploaded version language/pt_br-1.0.1.tar.gz differs from uploaded version main/communications-1.0.1.tar.gz differs from uploaded version main/fixed-0.7.1.tar.gz differs from uploaded version *** main/gsl-1.0.0.tar.gz differs from uploaded version but has same version main/image-1.0.1.tar.gz differs from uploaded version *** main/nnet-0.1.0.tar.gz differs from uploaded version but has same version main/octcdf-1.0.4.tar.gz differs from uploaded version main/odepkg-0.3.0.tar.gz differs from uploaded version main/optiminterp-0.2.2.tar.gz differs from uploaded version main/vrml-1.0.1.tar.gz differs from uploaded version *** main/audio-1.0.0.tar.gz differs from uploaded version but has same version main/econometrics-1.0.1.tar.gz differs from uploaded version main/general-1.0.1.tar.gz differs from uploaded version *** main/geometry-1.0.1.tar.gz differs from uploaded version but has same version *** main/linear-algebra-1.0.0.tar.gz differs from uploaded version but has same version *** main/miscellaneous-1.0.0.tar.gz differs from uploaded version but has same version main/optiminterp-0.2.2.tar.gz differs from uploaded version *** main/plot-1.0.0.tar.gz differs from uploaded version but has same version main/signal-1.0.1.tar.gz differs from uploaded version *** main/specfun-1.0.0.tar.gz differs from uploaded version but has same version main/statistics-1.0.1.tar.gz differs from uploaded version *** main/strings-1.0.0.tar.gz differs from uploaded version but has same version *** main/symbolic-1.0.1.tar.gz differs from uploaded version but has same version *** nonfree/arpack-1.0.0.tar.gz differs from uploaded version but has same version *** nonfree/spline-gcvspl-1.0.0.tar.gz differs from uploaded version but has same version *** Please fix the version numbering noted above before proceding make: *** [real_compare] Fejl 255 make: Forlader katalog '/home/sh/Dokumenter/octave-forge/octave-forge/packages' |
|
From: Michael G. <mic...@gm...> - 2007-05-23 20:37:00
|
On 5/23/07, dbateman <dba...@fr...> wrote: > > error: feval: the symbol `java_exit' is not valid as a function > > error: feval: the symbol `java_exit' is not valid as a function > > > > There is a problem with the java package.. I'm not going to chase this one. The problem is that the "autoload" for java_exit is wrong. Indeed, java_exit is autoloaded from __java__.oct. Typically, this happened to me recently, while the PKG_ADD file was not in the same dir as the oct-file (this is fixed now, but the octave version used for this test is 2.9.10). This might be the problem; this would also mean that any test based on an autoloaded function would fail. Michael |
|
From: dbateman <dba...@fr...> - 2007-05-23 20:51:44
|
Michael Goffioul-2 wrote: > > On 5/23/07, dbateman <dba...@fr...> wrote: >> > error: feval: the symbol `java_exit' is not valid as a function >> > error: feval: the symbol `java_exit' is not valid as a function >> > >> >> There is a problem with the java package.. I'm not going to chase this >> one. > > The problem is that the "autoload" for java_exit is wrong. Indeed, > java_exit is autoloaded from __java__.oct. Typically, this happened > to me recently, while the PKG_ADD file was not in the same dir as > the oct-file (this is fixed now, but the octave version used for this test > is 2.9.10). This might be the problem; this would also mean that any > test based on an autoloaded function would fail. > > Michael > > Ok, makes sense. Then this package should depend on 2.9.12 or greater.. I committed this change D. -- View this message in context: http://www.nabble.com/New-release--tf3794959.html#a10772545 Sent from the octave-dev mailing list archive at Nabble.com. |
|
From: Thomas T. <Tho...@gm...> - 2007-05-24 20:44:00
|
I downloaded the latest CVS 5 minutes ago. After that I added the
#include <endian.h>
to the top of connect.cc. Then I started once again from the beginning
make distclean...
> The is a bug in imrotate_Fourier.m. I fixed this in the CVS..
Fixed now ;)
Running test code for main/image-1.0.1.tar.gz and its dependencies
code/image-1.0.1/devel [tests 3 of 4 files]
---> success
code/image-1.0.1/inst [tests 30 of 84 files]
---> success
code/image-1.0.1/src [tests 5 of 13 files]
---> success
passes 232 out of 232 tests
> Can you try adding
>
> #include <endian.h>
>
> to the top of connect.cc and rebuild
Building main/parallel-1.0.1.tar.gz and its dependencies
export
INSTALLDIR="/Users/Thomas/Developement/forge/packages/sandbox/install/parallel-1.0.1";
make -C /var/tmp/oct-hgbN5w/parallel-1.0.1/src
mkoctfile -s connect.cc
Failed to build main/parallel-1.0.1.tar.gz!!!
Running test code for main/parallel-1.0.1.tar.gz and its dependencies
code/parallel-1.0.1/inst [tests 0 of 3 files]
---> success
code/parallel-1.0.1/src [tests 0 of 6 files]
---> success
passes 0 out of 0 tests
>>>>> Building main/statistics-1.0.1.tar.gz and its dependencies
>>>>> Running test code for main/statistics-1.0.1.tar.gz and its
>>>>> dependencies
>>>>> code/statistics-1.0.1/inst [tests 28 of 55 files]
>>>>> ---> passes 56 out of 58
>>>>> tests
>>>>> passes 56 out of 58 tests
Nearly David, I check it tomorrow together with the üatch from your second email.
Building main/statistics-1.0.1.tar.gz and its dependencies
Running test code for main/statistics-1.0.1.tar.gz and its dependencies
code/statistics-1.0.1/inst [tests 28 of 55 files]
---> passes 57 out of 58 tests
passes 57 out of 58 tests
Good night,
Thomas
|
|
From: Carlo de F. <car...@gm...> - 2007-05-22 12:32:13
|
On 5/22/07, David Bateman <Dav...@mo...> wrote: > The packages I have some sort of issue with can be checked with > > ./autogen.sh > ./configure > make all > cd packages > make checkpkg/extra/secs1d-0.0.1.tar.gz > make checkpkg/extra/secs2d-0.0.1.tar.gz This two packages are mine I'll have a look later today to see what are the issues. I also have done (but not yet committed) many modifications so probably I will change the version number too c. |
|
From: David B. <Dav...@mo...> - 2007-05-22 13:25:19
|
Carlo de Falco wrote: > On 5/22/07, David Bateman <Dav...@mo...> wrote: > >> The packages I have some sort of issue with can be checked with >> >> ./autogen.sh >> ./configure >> make all >> cd packages >> > > >> make checkpkg/extra/secs1d-0.0.1.tar.gz >> make checkpkg/extra/secs2d-0.0.1.tar.gz >> > > This two packages are mine I'll have a look later today > to see what are the issues. > I also have done (but not yet committed) > many modifications so probably I will change > the version number too > c. > > As there has never been an official release of these packages, changing the version number is not needed.. Regards David -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
|
From: David B. <Dav...@mo...> - 2007-05-22 13:26:24
|
Carlo de Falco wrote: > On 5/22/07, David Bateman <Dav...@mo...> wrote: > >> The packages I have some sort of issue with can be checked with >> >> ./autogen.sh >> ./configure >> make all >> cd packages >> > > >> make checkpkg/extra/secs1d-0.0.1.tar.gz >> make checkpkg/extra/secs2d-0.0.1.tar.gz >> > > This two packages are mine I'll have a look later today > to see what are the issues. > Maybe there are dependency issues on my installation. Haven't checked yet. I just flagged the packages that failed "make check" for me.. > I also have done (but not yet committed) > many modifications so probably I will change > the version number too > c. > > As there has never been an official release of these packages, changing the version number is not needed.. Regards David -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
|
From: dbateman <dba...@fr...> - 2007-05-23 05:28:33
|
S=C3=B8ren Hauberg wrote: >=20 >=20 > I've just started building octave-forge, but things break for me during= =20 > "make all". I get a large amount of >=20 > " > *** extra/civil-engineering-1.0.0.tar.gz differs from uploaded version=20 > but has same version > " >=20 > errors (for many (all?) packages, not just the civil-engineering). It=20 > finally exits with >=20 >=20 > " > *** Please fix the version numbering noted above before proceding > make: *** [real_compare] Fejl 255 > " >=20 > (Fejl =3D Error, it's Danish). It seems that newly created packages=20 > doesn't include a ChangeLog file, whereas the ones downloaded from SF > does. >=20 > S=C3=B8ren >=20 >=20 Are you perhaps trying to do the build on a machine without CVS access to Sourceforge? You must have CVS access to sourceforge from the source tree for "make all" to work.=20 D. --=20 View this message in context: http://www.nabble.com/New-release--tf3794959.= html#a10757747 Sent from the octave-dev mailing list archive at Nabble.com. |
|
From: Carlo de F. <car...@gm...> - 2007-05-23 07:02:43
|
On 23/mag/07, at 08:52, David Bateman wrote: > Or just do a "cvs login" before running "make all".... > > D. guglielmo.local $ cvs login cvs login: can only use `login' command with the 'pserver' method cvs [login aborted]: CVSROOT: :ext:cd...@oc...:/ cvsroot/octave Is not only ssh acces allowed to the sf repository? c. |
|
From: David B. <Dav...@mo...> - 2007-05-23 08:07:36
|
Carlo de Falco wrote: > > On 23/mag/07, at 08:52, David Bateman wrote: > >> Or just do a "cvs login" before running "make all".... >> >> D. > > guglielmo.local $ cvs login > cvs login: can only use `login' command with the 'pserver' method > cvs [login aborted]: CVSROOT: > :ext:cd...@oc...:/cvsroot/octave > > Is not only ssh acces allowed to the sf repository? > c. With ssh you have to have your key uploaded, so you shouldn't need "cvs login". I'm presuming that Soren is building on a machine with anonymous pserver access only to the CVS.. D. -- David Bateman Dav...@mo... Motorola Labs - Paris +33 1 69 35 48 04 (Ph) Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob) 91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax) The information contained in this communication has been classified as: [x] General Business Information [ ] Motorola Internal Use Only [ ] Motorola Confidential Proprietary |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X