🌐 Mailman 3 February 2003 - Pythonmac-SIG - python.org

one fallout of cr/lf code
by Magladry, Stephen March 21, 2003

March 21, 2003

It is nice to be able to run PC Python code unchanged thanks to the code change made to handle cr/lf within the Mac Python code. I have run into one problem through. The cr/lf is view as two separate lines. The following code, --test.py from a pc---- import sys <cr><lf> sys.blah<cr><lf> --test.py from a pc---- returns the following error: File "test.py", line 3 in ? AttributeError: 'module' object has no attribute 'blah' Okay I can deal with doubling of the line number, that's not much of a problem. The real problem is with line continuations. --test2.py from a pc---- x = x\ <cr><lf> +1<cr><lf> --test2.py from a pc---- This code is not able to execute. The line continuation "works" on the <lf>. The <cr> is seen as a new line. Now without a continuation, Mac Python isn't able to correctly process it. I can "fix" my problems my doing the following. --fixedTest2.py from a pc---- x = (x <cr><lf> +1)<cr><lf> --fixedTest2.py from a pc---- Since an expression gets started with the "(", line continuation come for free. This is fine and dandy for cross platform code that I own and work with. It might be more of a problem with Python code one would get from some PC user, though. Don't know if we can do anything about this. If not, consider this as a FYI for those people working in a cross platform environment. Sincerely, Stephen Magladry

4 5

Re: [Pythonmac-SIG] ObjC stuff
by Jack Jansen March 20, 2003

March 20, 2003

Recently, Steve Spicklemire <steve(a)spvi.com> said: > So it's in CVS somewhere? I'll do a check-out and see if I can find it. I don't know, Steven Majewski may know (I think he is the one who looked at it). If it isn't in the repository it probably should be, preferrably in the Modules directory (because it is useful for non-Mac users) but if there's objections against that we'll put it in the Mac/Modules dir. > I just ordered CW-6 today so I might be able to pitch in a bit. While very useful for other things (like MacPython development) you don't need it for OSX unix-Python development: this is all make and gcc-based (and you have those on your OSX developer disc). -- Jack Jansen | ++++ stop the execution of Mumia Abu-Jamal ++++ Jack.Jansen(a)oratrix.com | ++++ if you agree copy these lines to your sig ++++ www.oratrix.nl/~jack | see http://www.xs4all.nl/~tank/spg-l/sigaction.htm

5 4

sanity checks: building MacOS X app
by John Speno March 14, 2003

March 14, 2003

Dear MacPythonistas: I'm about to try bundling up my own python application for MacOS X 10.2. I'd like a little sanity check before I start. My question is: Is this going to work? I want to distribute this as a single application bundle (MyApp.app). I need Tkinter, so I know I'll need to have the TCL and Tk frameworks in my application. Of course it needs Python too. In addition, I need a version of the cyrus-sasl library compiled differently than the one that Apple includes with the OS. So, here's what I imagine it looking like: MyApp.app/ Info.plist MacOS/ MyApp - sh wrapper to launch ../Resources/myapp/myapp.py python -> ../Frameworks/Python.framework/.../python Frameworks/ TCL.framework/ Tk.framework/ Python.framework/ Resources/ myapp - my application modules lib/ libsasl2 Thanks!

8 14

State of Python on the Mac
by Nick Matsakis March 8, 2003

March 8, 2003

I've been following this list for the last month or so, and I must admit that as a long time Mac user and a new python programmer, I am still very confused about the state of Python on Mac OS X. I appreciate that a lot of work is going on to build a stable platform for the future, but the documentation is hopelessly out of date. Though the footer says "updated October 14, 2002", the documents at http://www.python.org/doc/current/mac/mac.html say "At the time of this writing Mac OS X had just been released as a Public Beta." Which makes me wonder if I can trust the rest of it at all. however, It doesn't seem like detailed documentation should be a priority now, since there is a lot of good Python documentation out there, and the documentation for Carbon, Cocoa, and other Mac OS X technologies is good and growing. Even Applescript is starting to get there, with the release of the O'Reilly book. Many programmers will be able to take that body of documentation and fill in the blanks. Instead, what I really want is a 1-2 page "state of Python on the Mac" written for a forward-looking programmer who is interested in using Python to script Mac OS X or write Carbon or Cocoa applications. The things it should provide are: 1. A bird's eye view of the APIs. The module list at http://www.python.org/doc/current/ is alphabetized, and thus random. I can make some reasonable intferences, but I really have no idea what modules "go together". If all goes well, a future Mac-python programmer will like have sets of modules that correspond roughly to "Cocoa" (pyobjc), "Carbon" (MacPython), "Applescript/OSA" (MacPython), and "standard python OS abstractions" (os, sys). Which modules belong where, which work well together, and which, if any, are deprecated? 2. A small description of the available and future python runtimes, and what to expect in the future. Apple included Python with 10.2. Why? Because it was in BSD 4.4? Because people asked for it? Has Apple made an official statement about its position on Python, or what versions of Python will be included with future versions of Mac OS X? (frameworks?) These are all important questions for the programmer who is trying to decide whether Python is appropriate for a given project. 3. A small description of the state of using Python as an OSA script. This is actually where _my_ primary interest lies. My impression of OSA is that it is supposed to allow different languages to be used to send and handle Apple Events. I hope that one day I would be able to install a simple component into the system and that suddenly I would be able to use Python as a full peer to Applescript; that is, anywhere that Applescript is used by applications and the system I could plug in Python. (My other impression is that the support may not yet be there from Apple, but that they are working on it, is this true?). 4. A solid answer to the question, "How can I help?" It's clear that documentation and other work is needed. Is the time appropriate to do that? How can documentation efforts be coordinated? I suppose this could be organized into a FAQ-style series of questions and answer, but I think it would be better as well-organized summary of the state of Python on the Mac. Regards, Nick Matsakis

12 19

Tkinter and the Help menu
by John Speno March 5, 2003

March 5, 2003

Platform: MacOS X 10.2.4, Python 2.3a2, Tcl and Tk 8.4.1 jaguar. Issue: In my application's Help menu, the first item is an extraneous 'x' which I didn't put there, and can't seem to get rid of. Actually, when I first switch to the application, each item in my Help menu is an 'x'. If I switch to another app then back, my menu items have their names back, and there's still the first mysterious 'x'. Anyone encounter that before? I suppose I can try the TOT build of Tcl/Tk... Thanks! #!/usr/local/bin/pythonw import Tkinter def hello(): print "hello" root = Tkinter.Tk() root.title('menubar test') menubar = Tkinter.Menu(root) helpmenu = Tkinter.Menu(menubar, name='help', tearoff=0) menubar.add_cascade(label="Help", menu=helpmenu, underline=0) addhelp = helpmenu.add_command addhelp(label='Window Help', underline=0, command=hello, state='normal') addhelp(label='Menu Help', underline=9, command=hello, state='normal') helpmenu.add_separator() addhelp(label='About Assignments', underline=0, command=hello) root.configure(menu=menubar) root.mainloop()

2 2

MacPython 2.3a2 on OSX 10.1
by Jack Jansen March 3, 2003

March 3, 2003

Pascal, John (and others), I think I've found the problem with running applets on 10.1. It seems that the new trick that Just invented, having the applet main program being a python (or shell) bootstrap script with a #! line, and using that script to fire up the real Python script, doesn't work under 10.1. I've tried a couple of #! scripts in .app bundles on 10.1, but they invariably give me an "Error -2857" on the console. So I guess this means that having a script as the main program in a .app bundle doesn't work under 10.1:-( I'm not sure how to proceed. We could scrap 10.1 support for MacPython, but that doesn't seem like a nice option. We could scrap Just's applet mods, but that I'm also not too excited about, the bundlebuilder stuff is a lot nicer than the buildtools hacks I used previously to create OSX applets. Moreover, bundlebuilder is also used for the PyObjC/applet-python-2.2 combo, and it's good to have only one way of doing things. We could replace the #! script by a true executable, but this would mean we'd have to store the boilerplate for that executable somewhere, and one of the nice things about the bundlebuilder solution is that it doesn't need any applet template anymore. Also, a true executable would still have to find out which script to run (although that could be coded in the .plist file, probably). -- Jack Jansen, <Jack.Jansen(a)cwi.nl>, http://www.cwi.nl/~jack If I can't dance I don't want to be part of your revolution -- Emma Goldman

7 9

error running test.py in 4Suite-0.12.0a3
by Joel Rodrigues March 1, 2003

March 1, 2003

Hi, Got another for you. I'm getting a weird error running test.py in 4Suite-0.12.0a3 [localhost:Ft/Share/Tests] mark% python test.py --help Traceback (most recent call last): File "test.py", line 9, in ? packages=['Lib', File "/usr/local/lib/python2.2/site- packages/Ft/Lib/TestSuite/__init__.py", line 45, in Test import TestSuite, Errors File "/usr/local/lib/python2.2/site- packages/Ft/Lib/TestSuite/TestSuite.py", line 16, in ? import TestModule, Tester File "/usr/local/lib/python2.2/site- packages/Ft/Lib/TestSuite/Tester.py", line 40, in ? curses.setupterm() _curses.error: setupterm: could not find terminal And are the demos in Python-2.2.2/Demo/curses supposed to work on Mac OS X v10.1.5, Python 2.2.2 ? This may the source of my problems : ncurses-5.2X.pkg, package built by Chris Roberts. I don't know where I got it. ANyone have a tip on an ncurses to download and for OS X ? - Joel

2 2

Re: [Pythonmac-SIG] MacPython 2.3a2 on OSX 10.1
by Daniel Lord Feb. 28, 2003

Feb. 28, 2003

I can verify that one can use and Applescript in an App bundle to call a python script in it. I created an Applescript Studio GUI as a wrapper for a recursive file renaming and moving Python script and it works very well. You find the path to the script in the bundle and call it as follows: (* Find path to python script *) set POSIXBasePath to POSIX path of (path to me) set scriptPath to (POSIXBasePath & scriptName) set theResult to do shell script ("/usr/local/bin/python " & "'" & scriptPath & "' " & cmdStr & fileArg) as string -- where cmdStr is your options and fileArg contains the command-line args -- this is how I call it, there are other ways Daniel Lord > From: Paul Berkowitz <berkowit(a)silcom.com> > Date: Fri Feb 28, 2003 08:52:50 US/Pacific > To: PythonMac <pythonmac-sig(a)python.org> > Subject: Re: [Pythonmac-SIG] MacPython 2.3a2 on OSX 10.1 > > > On 2/28/03 7:48 AM, "Kevin Ollivier" <kevino(a)tulane.edu> wrote: > >>> I've tried a couple of #! scripts in .app bundles on 10.1, but they >>> invariably give me an "Error -2857" on the console. >>> So I guess this means that having a script as the main program in a >>> .app bundle doesn't work under 10.1:-( >> >> Just a thought that crossed my mind - would an Applescript work? =) (I >> thought Applescript for 10.1 supported sending commands to the shell.) > > Yes, it does. 'do shell script "whatever"'. > > I can't recall if 'quoted form of POSIX path' came in with OS 10.2 (AS > 1.9) > or earlier. I think it was AS 1.8.3, which is a free update to OS > 10.1.5, > but will work with all versions of 10.1.x. Maybe even AS 1.8.1, which > shipped with OS 10.1.2. OS 10.1.0 shipped with AS 1.7, which is not a > good > version of AS, although I think it had 'do shell script' already. Best > would > be to require AS 1.8.3 and do all testing in OS 10.1 with that, but you > could even go back to 10.1.0 and just not use 'quoted form' but always > include all the "\"" or ' quotes yourself. > > -- > Paul Berkowitz > > > > > > >

1 0

aqua Tk
by Russell E Owen Feb. 27, 2003

Feb. 27, 2003

> > >>Although, I highly recommend NOT using aquaTk. >> > >> >I'll bite: Why not? >> >> It's unstable (something of a late alpha quality) and ugly on OS X, you >> should use PyObjC if possible, if not, try wxWindows (if the GUI has to >> be cross-platform) or something.. tkinter is kind of old and convoluted > > too... I've personally found aqua Tk 8.4.1 (the current version last I checked) to be just fine. It has a few cosmetic bugs that I've reported (insertion cursor sometimes not shown in a right justified entry widget, lines on canvasses are off by one pixel relative to unix (which was also true in MacOS 9)), but my telescope GUI seems perfectly stable. I don't mean to dismiss your own experience -- my code may well not tickle he bugs you've seen. However, I would be interested in knowing of any serious bugs. And did you see them in 8.4.1 (aqua Tk 8.4 was definitely buggy)? -- Russell

2 1

Re: [Pythonmac-SIG] Slow loading of modules
by Jack Jansen Feb. 24, 2003

Feb. 24, 2003

On zaterdag, feb 22, 2003, at 12:30 Europe/Amsterdam, Doug Wyatt wrote: > Try this: > > /python.exe -c "import Carbon.AE" & ; sample python.exe 10 10 > > and then look at the sample log written to /tmp Doug, thanks a lot!!!! I looked at this, and it turns out most of the time is spent in _dyld_link_module, and specifically in _resolve_undefineds and _relocate_modules_being_linked. On a hunch I disabled prebinding for the main executable, which I had enabled only a couple of weeks ago. The results are staggering: - With prebinding: % time ./python.exe -c "pass" 0.130u 0.130s 0:00.62 41.9% 0+0k 24+3io 0pf+0w % time ./python.exe -c "import _AE" 8.140u 0.170s 0:08.54 97.3% 0+0k 8+4io 0pf+0w - Without prebinding: % time ./python.exe -c "pass" 0.100u 0.130s 0:00.28 82.1% 0+0k 0+0io 0pf+0w % time ./python.exe -c "import _AE" 1.130u 0.110s 0:01.39 89.2% 0+0k 0+1io 0pf+0w I definitely do remember a speedup for the initial empty startup (I did actually test that -prebind made a difference:-), so the fact that even that is now worse may be an artifact of changes in site.py or something. It appears that a program that does a lot of dynamic loading of code is worse off when linked prebound. Does anyone have a good explanation for this, or should I file an Apple bug report (even only suggesting that they document this)? -- - Jack Jansen <Jack.Jansen(a)oratrix.com> http://www.cwi.nl/~jack - - If I can't dance I don't want to be part of your revolution -- Emma Goldman -

3 2