Hi, I have now created a little ugly hack in the relax main 1.3 line to work around the Mac Application problem with the missing Carbon.Evt.TickCount error message (due to Python version issues). This is as I described at https://mail.gna.org/public/relax-users/2012-04/msg00021.html, the command line argument emulation has been turned off relax now works around this. I have created a special relax application DMG file for testing at http://download.gna.org/relax/. The temporary file is called relax-1.3_tickcount_fix.Darwin.dmg (with a md5sum of d2b2c4455ea9783023d81e096e22072f). Could I ask anyone who is suffering from, or who can trigger, this TickCount problem if you could check and see if the problem is now eliminated?
Cheers, Edward On 5 April 2012 10:33, Edward d'Auvergne <[email protected]> wrote: > Hi Ben, > > Thank you for that info, that will really help in solving some of > these problems! I never though that one of the issues is that of the > machine architecture. This is obviously a multi-layer, i.e. not just > one failure point, problem with the py2app script for converting > Python programs/scripts into full Mac Applications > (http://wiki.python.org/moin/MacPython/py2app, > http://svn.pythonmac.org/py2app/py2app/trunk/doc/index.html). I might > have to approach this in a number of ways, some fixes will probably > allow relax to run for some Mac users while others will not be so > lucky. In the first attack, I will try to eliminate this TickCount > problem. This seams to be a problem with program argument emulation > on the Mac with certain Python versions. A solution is posted at > http://stackoverflow.com/questions/3461983/evt-tickcount-not-found-with-python2-6-on-osx-10-6-3, > where the suggested code is: > > APP = ['MyApp.py'] > OPTIONS = { > 'argv_emulation': False, > ... > } > > setup( > app=APP, > options={'py2app': OPTIONS}, > setup_requires=['py2app'], > ) > > In relax's py2app setup.py script, the 'argv_emulation' is currently > set to True (see > http://svn.gna.org/viewcvs/*checkout*/relax/1.3/setup.py?revision=15481&content-type=text%2Fplain). > relax currently requires arguments, but maybe I can hack the code to > catch this and skip the argument processing. But only for this > situation as the prompt/scripting UI modes will be important for some > users. > > This fix doesn't solve the problem of the py2app created relax app not > using the bundled Python version and internal Python modules, so some > users will still be affected by that. For some people this is not an > issue while for others it causes relax to fail. > > A third problem might be in the wxPython code, as was seen and > reported by Jack Howarth when creating the Fink distribution of relax > (see the threads starting at > https://mail.gna.org/public/relax-users/2012-03/msg00053.html and > https://mail.gna.org/public/relax-users/2012-04/msg00003.html, and > http://pdb.finkproject.org/pdb/browse.php?name=relax). This will > probably also affect those who install relax from the sources > distribution file > (http://www.nmr-relax.com/download.html#Source_code_release). The > problem might require a workaround inside the relax GUI code. I will > try to tackle each problem one by one, and hopefully as each is solved > more and more users will have luck in running relax in GUI mode on the > Mac. > > Cheers, > > Edward > > > P. S. For those users who can currently run the relax application in > GUI mode from the DMG file, these fixes should not affect you. > > > > On 4 April 2012 18:58, Ben Eisenbraun <[email protected]> wrote: >> Hi Edward, >> >> I am seeing the same issues reported by others. >> >> $ open /programs/i386-mac/relax/1.3.14/relax.app >> >> Output in Console.app: >> >> 4/4/12 12:15:26.639 PM [0x0-0x18a28a1].com.nmr-relax.relax: Traceback (most >> recent call last): >> 4/4/12 12:15:26.639 PM [0x0-0x18a28a1].com.nmr-relax.relax: File >> "/nfs/programs/i386-mac/relax/1.3.14/relax.app/Contents/Resources/__boot__.py", >> line 126, in <module> >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: >> _argv_emulation() >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: File >> "/nfs/programs/i386-mac/relax/1.3.14/relax.app/Contents/Resources/__boot__.py", >> line 124, in _argv_emulation >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: >> _get_argvemulator().mainloop() >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: File >> "/nfs/programs/i386-mac/relax/1.3.14/relax.app/Contents/Resources/__boot__.py", >> line 63, in mainloop >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: stoptime = >> Evt.TickCount() + timeout >> 4/4/12 12:15:26.640 PM [0x0-0x18a28a1].com.nmr-relax.relax: AttributeError: >> 'module' object has no attribute 'TickCount' >> 4/4/12 12:15:26.715 PM relax: relax Error >> >> This is a problem related to Carbon support in 32-bit vs 64-bit Python. I >> can reproduce this on the command line in 10.6 and 10.7 like so: >> >> $ type python >> python is hashed (/usr/bin/python) >> $ python >> Python 2.7.1 (r271:86832, Jun 16 2011, 16:59:05) >> [GCC 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2335.15.00)] on >> darwin >> Type "help", "copyright", "credits" or "license" for more information. >>>>> from Carbon import Evt >>>>> Evt.TickCount() >> Traceback (most recent call last): >> File "<stdin>", line 1, in <module> >> AttributeError: 'module' object has no attribute 'TickCount' >> >> And it's because the 64-bit Python doesn't have full Carbon support: >> >> $ export VERSIONER_PYTHON_PREFER_32_BIT=true >> $ python >> Python 2.7.1 (r271:86832, Jun 16 2011, 16:59:06) >> [GCC 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2335.15.00)] on >> darwin >> Type "help", "copyright", "credits" or "license" for more information. >>>>> from Carbon import Evt >>>>> Evt.TickCount() >> 278493166 >> >> Similar thread here: >> >> http://stackoverflow.com/questions/3461983/evt-tickcount-not-found-with-python2-6-on-osx-10-6-3 >> >> I can run the relax binary directly if I force 32-bit: >> >> $ arch -i386 relax.app/Contents/MacOS/relax >> >> And I can get the relax.app bundle to work if I lipo the relax binary: >> >> $ mv relax relax.orig >> $ lipo -thin i386 -output relax relax.orig >> >> I tried various other ways of forcing 32-bitness for Python: lipo'ing the >> python binaries, export VERISONER_PYTHON_PREFER_32_BIT=true, >> LSArchitecturePriority in the Info.plist without effect. >> >> -ben >> >> -- >> | Ben Eisenbraun >> | SBGrid Consortium | http://sbgrid.org | >> | Harvard Medical School | http://hms.harvard.edu | _______________________________________________ relax (http://www.nmr-relax.com) This is the relax-users mailing list [email protected] To unsubscribe from this list, get a password reminder, or change your subscription options, visit the list information page at https://mail.gna.org/listinfo/relax-users
