🌐 Mailman 3 December 2009 - Pythonmac-SIG - python.org

Re: [Pythonmac-SIG] Bundlebuilder--why remove it?
by has Dec. 12, 2009

Dec. 12, 2009

Christopher Barker wrote: > It's open-source, it's only going to get done if someone that needs it rolls up his/her sleeves and does it -- which may mean learning more about the nitty gritty details of OS-X. Yep. If you want something done right, sometimes you just gotta step up and do it yourself. It's that or large cash bribes to get someone else to do it; take your pick. > My suggestion to Charles, and Kevin, (and me!) is to try to find some time to dig into py2app -- I think once you grok the code structure, you can probably find the source of the errors, and then, if you can't figure out a solution, Ronald or Ned (or??) might be able to help out, without them having to do all the work -- i.e. use the experts only for the part where their expertise is critical. I'd go further, and suggest pulling py2app to pieces. Creating an executable stub is a separate task to bundling project files is a separate task to locating and bundling third-party module dependencies is a separate task to bundling a Python framework (full or stripped). Some of these parts have more intrinsic value than others (e.g. identifying module dependencies), particularly as they can be used cross-platform and so will be easier to muster maintainers. Rather than revive the moribund bundlebuilder, put the time you'd spend on that into bringing the Xcode-Python templates up to scratch. The executable stub is a piece of crap that imports the first Python framework it finds, as opposed to the version you built with; fixing this should not be too difficult if you know some C/ObjC (dynamically binding framework functions is not that hard - I've done it myself - and you can borrow the code from py2app). Once you've got Xcode building Python-based .app bundles that don't suck, you can look at providing dependency bundling as an extra build phase. Personally, I like py2app and I'm not a huge fan of Xcode for Python development, but the first rule of sustainable OSS development has to be: when you have limited time, knowledge and manpower at your disposal, invest it on functionality that can't be leveraged from elsewhere. Xcode is not ideal, but it's good enough. In addition, Xcode's privileged position in OS X development makes it the first stopping point for newcomers to Python+PyObjC development, so targeting it will help market Python for Mac application development and strengthen the community as a whole. HTH has p.s. Re. the original subject line, kicking bundlebuilder and the rest of the decaying/decayed Mac-specific libraries out of Python's stdlib was completely the right decision for Python's overall health. -- Control AppleScriptable applications from Python, Ruby and ObjC: http://appscript.sourceforge.net

2 1

[ann] py-appscript 0.21.1
by has Dec. 12, 2009

Dec. 12, 2009

Hi all, Just posted a bug-fix release to PyPI that corrects a problem when unpacking Boolean values on Python 3.x. (The 1.0 release date is pushed back accordingly.) Please go beat harder on it. :) has -- Control AppleScriptable applications from Python, Ruby and ObjC: http://appscript.sourceforge.net

1 0

Re: [Pythonmac-SIG] deconstruct a py-appscript reference
by has Dec. 12, 2009

Dec. 12, 2009

Mark Sanvitale wrote: > I am new to both python and to appscript (I took up both at the same time). I am curious about what I think is an internal detail of the appscript module and its classes. Given my newbie status, I apologize in advance if my terminology for appscript things strays from the standard. > > I have a reference returned by calling get() on some other reference (e.g. asking an iphoto album for its child list). The reference I get back is fully resolved to a unique item (i.e. of the form app(u'/Applications/iPhoto.app').albums.ID(4294967326) and not involving a by-name lookup or query). Now, I can use the returned reference to get() its properties, however, if the property I am after is "id" then it would be wasteful (i.e. extra IPC) to perform a get() for the id property because I can see it is a component of the reference I already have. > > Thus, my question is how can I get the id directly out of such a reference? Presumably, via some means more elegant than parsing a string version of the reference. (So far, I have gotten as far as poking around myRef.AS_appdata and myRef.AS_aemreference but have not yet found the magic word.) If deep appscript juju is your thing, the aem package's Query subclasses provide a visitor method, AEM_resolve, for walking the entire reference. (The appscript.referencerenderer module uses it.) However, unless you're sending many thousands of events then asking the application for the object's id is unlikely to be a performance bottleneck. Personally I'd just ask the application, and maybe run a Python profiler if I think I'm still not getting my money's worth. If you are sending many thousands of events, there are probably other optimisation techniques you should be using anyway. Despite superficial appearances, Apple event IPC is a combination of RPC+queries; trying to apply object-oriented idioms to appscript code when dealing with large data sets is a common cause of poor performance. See chapter 13 of the appscript manual for more information, and post back here if you need more advice. HTH has -- Control AppleScriptable applications from Python, Ruby and ObjC: http://appscript.sourceforge.net

1 0

Bundlebuilder--why remove it?
by Kevin Walzer Dec. 11, 2009

Dec. 11, 2009

After spending some time getting bundlebuilder to work with my 64-bit programs, I have begun to wonder: what's the rationale for removing it from Python 3.X? Its Carbon dependencies can be eliminated by simply removing the "import argvemulator" statement, so (as far as I can see) there is no serious technical impediment to leaving it in the Python library. Bundlbuilder doesn't really argvemulation anyway, since hooks for this exist in all the major GUI libraries (Tk, wxPython, and certainly PyObjC). And while bundlebuilder is a less robust tool than py2app, it is nonetheless capable enough: I'm shipping two commercial applications that make use of it. In fact, its simplicity (a single module) make it easier to dig into and patch, something I can't do with py2app. I wouldn't have been able to release my applications without it. What's the best way to keep bundlebuilder available for Python 3.x? Submit a feature request at the bug tracker? Or separate it out, and submit a PyPi project? --Kevin -- Kevin Walzer Code by Kevin http://www.codebykevin.com

4 12

deconstruct a py-appscript reference
by Mark Sanvitale Dec. 11, 2009

Dec. 11, 2009

I am new to both python and to appscript (I took up both at the same time). I am curious about what I think is an internal detail of the appscript module and its classes. Given my newbie status, I apologize in advance if my terminology for appscript things strays from the standard. I have a reference returned by calling get() on some other reference (e.g. asking an iphoto album for its child list). The reference I get back is fully resolved to a unique item (i.e. of the form app(u'/Applications/iPhoto.app').albums.ID(4294967326) and not involving a by-name lookup or query). Now, I can use the returned reference to get() its properties, however, if the property I am after is "id" then it would be wasteful (i.e. extra IPC) to perform a get() for the id property because I can see it is a component of the reference I already have. Thus, my question is how can I get the id directly out of such a reference? Presumably, via some means more elegant than parsing a string version of the reference. (So far, I have gotten as far as poking around myRef.AS_appdata and myRef.AS_aemreference but have not yet found the magic word.) Thanks. Mark Sanvitale

2 1

Newbie question: py2app does only include main .py file, not the rest.
by Ron Arts Dec. 9, 2009

Dec. 9, 2009

Hi, I am porting an app that was initially written for Windows and Linux to OSX. For Linux an .rpm is created, for Windows I used a py2exe setup.py script. I adapted that script for OSX, and it seems to do a lot of work, generates an App Bundle, but in the Contents/Resources directory only the main .py file is created, not all other .py files from the application. If I copy those in manually I get: ImportError: No module named pkg_resources. Now I'm at a loss how to continue. What should I do next? Thanks, Ron

2 1

Bundlebuilder and four-way universal binaries: alternative to py2app
by Kevin Walzer Dec. 9, 2009

Dec. 9, 2009

Hi all, I wanted to report that I've successfully built a four-way universal binary of one of my Python applications using bundlebuilder in Python 2.6.2. I previously had problems with this because I got errors about argvemulator not being supported (apparently a Carbon function that doesn't exist in 64-bit mode). Here are the steps I've taken: 1. In your setup/makeapplication.py script, don't import argvemulation. 2. When you run the setup script, make sure you call python2.6-all as the Python interpreter--this will hit the 64-bit and 32-bit versions. My call looks like this: python2.6-all makeapplication.py build And that's it. I realize that bundlebuilder is going away in Python 3.x, and I have no idea if it would work if you simply separated out the bundlebuilder.py module and ran it with your own Python 3.x applications. I also understand that bundlebuilder is quirky and a less graceful solution than py2app: you have to do some trial and error to get all the packages bundled in your app. However, until py2app plays nice with four-way universal binaries, bundlebuilder will be a good short- and mid-term solution for me. Just wanted to share, Kevin -- Kevin Walzer Code by Kevin http://www.codebykevin.com

1 0

py2app/PyQt/64-bit/Mach-O
by Xiao Yu Dec. 8, 2009

Dec. 8, 2009

Hi, I think this problem might have been previously discussed but I'm not too sure what the solution is. I'm attempting to build my PyQt4 application. I'm receiving an error here: File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/py2app/build_app.py", line 589, in _run self.run_normal() File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/py2app/build_app.py", line 660, in run_normal self.create_binaries(py_files, pkgdirs, extensions, loader_files) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/py2app/build_app.py", line 768, in create_binaries mm.mm.run_file(runtime) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachOGraph.py", line 66, in run_file m = self.createNode(MachO, pathname) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachOStandalone.py", line 23, in createNode res = super(FilteredMachOGraph, self).createNode(cls, name) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/altgraph/ObjectGraph.py", line 148, in createNode m = cls(name, *args, **kw) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachO.py", line 63, in __init__ self.load(file(filename, 'rb')) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachO.py", line 78, in load self.load_header(fh, 0, size) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachO.py", line 108, in load_header hdr = MachOHeader(self, fh, offset, size, magic, hdr, endian) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachO.py", line 148, in __init__ self.load(fh) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/MachO.py", line 161, in load header = self.mach_header.from_fileobj(fh, **kw) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/ptypes.py", line 44, in from_fileobj return cls.from_str(f.read(cls._size_), **kw) File "/opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/ptypes.py", line 48, in from_str return cls.from_tuple(struct.unpack(endian + cls._format_, s), **kw) error: unpack requires a string argument of length 32 > /opt/local/Library/Frameworks/Python.framework/Versions/2.6/lib/python2.6/site-packages/macholib/ptypes.py(48)from_str() -> return cls.from_tuple(struct.unpack(endian + cls._format_, s), **kw) Any solutions? I have qt4-mac 4.5.3 python26 @2.6.4_0+darwin py26-sip @4.9.3_0 py26-pyqt4 @4.6.2_0 py26-py2app-devel @0.4.4_0 py26-macholib @1.2.1_2 Regards, Xiao

2 1

Installing Python 2.6.4 alongside stock python 2.6.1 on SnowLeopard
by Erik van Blokland Dec. 7, 2009

Dec. 7, 2009

Hi, sorry to bother, but I've run into a spot of trouble in an area I'm not familiar with. (Ronald, this is a follow up of the post I sent yesterday, I'm sorry for the moving target. I'm stuck.) I installed python 2.6.4 on Snow Leopard 10.6.2. I used the prepackaged .dmg installer from python.org. The new 2.6.4: [eBook:pyobjc/trunk/launch] erik% python Python 2.6.4 (r264:75821M, Oct 27 2009, 19:48:32) The stock python 2.6.1 is also still there: [eBook:pyobjc/trunk/launch] erik% /usr/bin/python Python 2.6.1 (r261:67515, Jul 7 2009, 23:51:51) With casual testing the new 2.6.4 seems to work well. But when installing other libraries I get very peculiar crashes. Installing Twisted and Pyobjc (both from source, not easy_install) end in a similar way with lipo complaining and references to a 10.4 sdk: /Developer/SDKs/MacOSX10.4u.sdk. The /Developer/SDKs dir contains a 10.6 sdk. But I am not familiar with what happens or should happen in this place. I posted a question about this on the twistd list earlier, but now I see the same thing happen with trying to install pyobjc, I suspect something in my installation has gone wrong. My next step seems to be to reinstall Snow Leopard. Any pointers are welcome. Could it be related to 32 / 64 bit issues? Regards, Erik Tail of the pyobjc install log: building 'launch._launch' extension gcc-4.0 -arch ppc -arch i386 -isysroot /Developer/SDKs/MacOSX10.4u.sdk -fno-strict-aliasing -fno-common -dynamic -DNDEBUG -g -O3 -I/Library/Frameworks/Python.framework/Versions/2.6/include/python2.6 -c Modules/_launch.c -o build/temp.macosx-10.3-fat-2.6/Modules/_launch.o In file included from /Library/Frameworks/Python.framework/Versions/2.6/include/python2.6/unicodeobject.h:4, from /Library/Frameworks/Python.framework/Versions/2.6/include/python2.6/Python.h:85, from Modules/_launch.c:1: /Developer/SDKs/MacOSX10.4u.sdk/usr/include/stdarg.h:4:25: error: stdarg.h: No such file or directory In file included from /Library/Frameworks/Python.framework/Versions/2.6/include/python2.6/unicodeobject.h:4, from /Library/Frameworks/Python.framework/Versions/2.6/include/python2.6/Python.h:85, from Modules/_launch.c:1: /Developer/SDKs/MacOSX10.4u.sdk/usr/include/stdarg.h:4:25: error: stdarg.h: No such file or directory lipo: can't figure out the architecture type of: /var/tmp//ccNyhKCK.out error: command 'gcc-4.0' failed with exit status 1 [eBook:pyobjc/trunk/launch] erik%

2 2

Re: [Pythonmac-SIG] Can't build PIL on Snow Leopard
by Xiao Yu Dec. 7, 2009

Dec. 7, 2009

There is no problem at all when I use Apple Python with pretty much everything. Except when I have to distribute it with py2app. A separate Python is essential for me. I could use MacPort or Fink but that would put me a few versions behind the latest release. I'd really like to just use Python from python.org and add all the libraries I need but I can't seem to compile any libraries for this Python. They all end with a similar error. I'm not sure what I'm missing. It gives the same error when I try to egg this too. Regards, Xiao > i have a PIL binary for python2.6.1 on Snow Leopard. I use the Apple-native python, and i haven't run the module through any tests to verify it's soundness, but just doing "import Image" in python works just fine... > > let me know if anyone would like it, or would like to discuss. I can always be found on #utahpython on IRC under the name mgeary. > > best, > > michael geary > > > On Dec 3, 2009, at 5:52 PM, Christopher Barker wrote: > > > First: > > > > Does anyone have a PIL binary for Python2.6.4? That would be the easiest way to get it? > > > > I built one for Frederik last spring, but I don't know what happened to it, nor do I remember whether it was for py2.5 or py2.6 > > > > It's not too hard to do -- at least with 10.5 machine -- I have a 10.4.11 PP now, and macports does not build Universal right at this point :-( > > > > Xiao Yu wrote: > >> I apologize for not having caught onto the nuance of differences between the Apple system Python, python.org's Python and MacPort's Python yet. I have installed python.org's Python 2.6.4. I however have trouble building PIL on it. > >> lipo: can't open input file: /var/tmp//ccmUAK0V.out (No such file or directory) > >> error: Setup script exited with error: command 'gcc-4.0' failed with exit status 1 > >> I am being told by the folks at comp.lang.python that the problem is because of missing libraries. Could someone please instruct me how to proceed? I couldn't google the proper procedures. My ultimate goal is to get my PyQt-PIL application working again. > > > > That is likely. If you look at the error messages that came before, you may find more details, but PIL does depend in a couple other libs that Apple does not provide. > > > > You will find a list of these libs in the PIL docs, and probably discussion about Mac-specific issue in the archives of this list and assorted Wikis -- do some goggling. I think you need libjpeg and libfreetype, though freetype may now come with OS-X 10.6 > > > > The easiest way to get them is to use macports (or maybe fink). > > > > If you just need it to work for you, then it's pretty easy. Use macports to build libjpeg, and make sure the PIL build can find it, and away you go. > > > > If you want to re-distribute, then you'll want to build the libd Universal, which macports shoudl be able to do with: > > > > port install libjpeg _universal > > > > or something like that. Then you want make sure that PIL finds only the static libs -- I think macports builds both, so you may need to go delete the *.dylib versions so that they won't be found -- a kludge, but gcc really tries hard to use dynamic libs. > > > > Sorry I don't have time to do more right now! > > > > -Chris > > > > > > > > > > -- > > Christopher Barker, Ph.D. > > Oceanographer > > > > Emergency Response Division > > NOAA/NOS/OR&R (206) 526-6959 voice > > 7600 Sand Point Way NE (206) 526-6329 fax > > Seattle, WA 98115 (206) 526-6317 main reception > > > > Chris.Barker at noaa.gov > > _______________________________________________ > > Pythonmac-SIG maillist - Pythonmac-SIG at python.org > > http://mail.python.org/mailman/listinfo/pythonmac-sig >

3 2