pyobjc-dev — CVS commit notices go here

Re: [Pyobjc-dev] PB templates currently broken

[Jack J. <Jac...@or...>]

On vrijdag, mei 2, 2003, at 18:38 Europe/Amsterdam, Dinu Gherman wrote:

> Bill Bumgarner:
>
>> The project builder templates are currently a bit broken -- they 
>> don't create projects that run "out of t he box".   It looks to be do 
>> to the namespace cleanup that happened over the past weeks/months.
>>
>> I'm not going to have time to address this before I return from 
>> California next Wednesday -- if someone can beat me to it, that would 
>> be great [and would be good to do before the 0.9 release].
>
> Well, I'm not beating anybody. I just looked into the website again
> and found that 0.8 was soon released five months ago (as it claims,
> on 2002-12-19). So, given the many changes inbetween and given that
> EuroPython is about 50 days away I'd really like to see 0.9 come out
> before June and, ideally with some time left for people to use it
> and adapt their apps, if needed, before going there, but YMMV...
> I'd even swallow some light bugs in 0.9, but the templates should
> be working.

I second this. I know, it's easy to second something if you're not 
willing to actually
do the dirty work, but still:-) Actually, I *am* willing to do some 
work. If there
are only one or two showstoppers: please let me know and I'll see if I 
can find time.
If someone can give me a recipe to see the problem, that is (Bill's 
description
is a bit too vague).

Ideally I would like to see 0.9 Real Soon Now, as in early next week. 
If that
happens I will delay the announcement of MacPython-2.3b1 so that PyObjC 
0.9
can be installed through the package manager. I think we have a really 
cool
combo, then, and something we can get some airplay for, to whet people's
appetite so we'll blow them away when 2.3 and 1.0 come out later.

But I'm off on a holiday after next week, and as the 2.3b1 is the first
one that's going to get exposure (and, hence, bug reports) I really 
want to
advertise it next week.

And lets try for a fast track to 1.0. Don't forget 1.0 doesn't have to 
be perfect,
after all "it's only a 1.0 release". And we shouldn't forget about 
"release
early, release often". We don't want (say) the Ruby people to come up 
with
something before us that will eat the publicity and turn us into
also-runners.
--
- Jack Jansen        <Jac...@or...>        
http://www.cwi.nl/~jack -
- If I can't dance I don't want to be part of your revolution -- Emma 
Goldman -

Re: [Pyobjc-dev] PB templates currently broken

[Dinu G. <gh...@da...>]

Bill Bumgarner:

> The project builder templates are currently a bit broken -- they don't 
> create projects that run "out of t he box".   It looks to be do to the 
> namespace cleanup that happened over the past weeks/months.
>
> I'm not going to have time to address this before I return from 
> California next Wednesday -- if someone can beat me to it, that would 
> be great [and would be good to do before the 0.9 release].

Well, I'm not beating anybody. I just looked into the website again
and found that 0.8 was soon released five months ago (as it claims,
on 2002-12-19). So, given the many changes inbetween and given that
EuroPython is about 50 days away I'd really like to see 0.9 come out
before June and, ideally with some time left for people to use it
and adapt their apps, if needed, before going there, but YMMV...
I'd even swallow some light bugs in 0.9, but the templates should
be working.

Regards,

Dinu

--
Dinu C. Gherman
......................................................................
"There is more to life than increasing its speed." (Mahatma Ghandi)

[Pyobjc-dev] PB templates currently broken

[Bill B. <bb...@co...>]

The project builder templates are currently a bit broken -- they don't 
create projects that run "out of t he box".   It looks to be do to the 
namespace cleanup that happened over the past weeks/months.

I'm not going to have time to address this before I return from 
California next Wednesday -- if someone can beat me to it, that would 
be great [and would be good to do before the 0.9 release].

b.bum

Re: [Pyobjc-dev] NSSavePanel causing SIGBUS 10

[Just v. R. <ju...@le...>]

Greg Hamilton wrote:

> Hello,
> 
> We're trying to display a standard NSSavePanel save file sheet.
> The save sheet is displayed.
> The Cancel button works.
> Save causes a SIGBUS 10.
> Any clues ?
> 
>      def saveAs_(self, sender):
>          sp = NSSavePanel.savePanel()
>           
>
sp.beginSheetForDirectory_file_modalForWindow_modalDelegate_didEndSelect 
> or_contextInfo_(
>           '~/Documents',
>           None,
>           self.mainWindow,
>           self,
>           'didEndSheet:returnCode:contextInfo:',
>           0)
>          print "end of save as"
> 
>      def didEndSheet_returnCode_contextInfo_(self, sheet, return_code,  
> contextInfo):
>          print 'Did End'
>          if return_code == NSOKButton:
>              self.filename = sheet.filename()
>              print "saving %s items as %s" % (len(self.sched),  
> self.filename)
>              self.sched.save(self.filename)

You need to specify the exact signature of your didEndSheet_etc method:
returnCode is an int argument and PyObjC can't guess that. After your
method definition, add:

    didEndSheet_returnCode_contextInfo_ =
objc.selector(didEndSheet_returnCode_contextInfo_, signature="v@:@ii")

(I'm not sure I got this right as I don't know where that signature
syntax is defined and I have no time to actually run the code.)

Just

[Pyobjc-dev] NSSavePanel causing SIGBUS 10

[Greg H. <gr...@ob...>]

Hello,

We're trying to display a standard NSSavePanel save file sheet.
The save sheet is displayed.
The Cancel button works.
Save causes a SIGBUS 10.
Any clues ?

     def saveAs_(self, sender):
         sp = NSSavePanel.savePanel()
          
sp.beginSheetForDirectory_file_modalForWindow_modalDelegate_didEndSelect 
or_contextInfo_(
          '~/Documents',
          None,
          self.mainWindow,
          self,
          'didEndSheet:returnCode:contextInfo:',
          0)
         print "end of save as"

     def didEndSheet_returnCode_contextInfo_(self, sheet, return_code,  
contextInfo):
         print 'Did End'
         if return_code == NSOKButton:
             self.filename = sheet.filename()
             print "saving %s items as %s" % (len(self.sched),  
self.filename)
             self.sched.save(self.filename)

Greg Hamilton

Re: [Pyobjc-dev] scary problem resolved!

[Ronald O. <ous...@ci...>]

On Saturday, Apr 26, 2003, at 00:44 Europe/Amsterdam, Jack Jansen wrote:

>
> On vrijdag, apr 25, 2003, at 16:48 Europe/Amsterdam, Just van Rossum 
> wrote:
>> Then the final clue came in: Bill had not seen the problem when _he_
>> built the app, but did reproduce it with the one _I_ built. So yes, it
>> had something to do with my setup (as I sortof suspected), and all 
>> along
>> I overlooked one thing: I didn't have the exact same version of libffi
>> as he did. Fixing that made the problem go away. Yay!
>
> Can we create a regression test for this? Is it worth it?

Maybe we can, but I don't think it is worth it unless this is very 
easy. I think this problem might be related to the warning we used to 
get when linking with libffi (about relocations in a TEXT section).

Ronald (yes, he's back)

Re: [Pyobjc-dev] scary problem resolved!

[Jack J. <Jac...@or...>]

On vrijdag, apr 25, 2003, at 16:48 Europe/Amsterdam, Just van Rossum 
wrote:
> Then the final clue came in: Bill had not seen the problem when _he_
> built the app, but did reproduce it with the one _I_ built. So yes, it
> had something to do with my setup (as I sortof suspected), and all 
> along
> I overlooked one thing: I didn't have the exact same version of libffi
> as he did. Fixing that made the problem go away. Yay!

Can we create a regression test for this? Is it worth it?
--
- Jack Jansen        <Jac...@or...>        
http://www.cwi.nl/~jack -
- If I can't dance I don't want to be part of your revolution -- Emma 
Goldman -

[Pyobjc-dev] scary problem resolved!

[Just v. R. <ju...@le...>]

First off, thanks a lot to everyone who sent me their test results. It
seems all G4 machines were affected, but not G3's.

Then the final clue came in: Bill had not seen the problem when _he_
built the app, but did reproduce it with the one _I_ built. So yes, it
had something to do with my setup (as I sortof suspected), and all along
I overlooked one thing: I didn't have the exact same version of libffi
as he did. Fixing that made the problem go away. Yay!

Just

Re: [Pyobjc-dev] Re: Talk status for EPC 2003?

[Just v. R. <ju...@le...>]

Michael Hudson wrote:

> > I can think about giving an overview talk about Cocoa if nobody
> > else wants to. In that case I might be teasing this list for an
> > exchange of ideas and input, though... ;-)
> 
> I've been emailing with Just about a more user oriented talk, but less
> is settled here I think (like for instance which track it belongs in).

An exchange of ideas for a talk would be a great idea regardless. Dinu
has much more experience with Cocoa that I have, so he should be more
qualified to do a general Cocoa talk. Lets figure something out.

Just

[Pyobjc-dev] Re: Talk status for EPC 2003?

[Michael H. <mw...@py...>]

Dinu Gherman <gh...@da...> writes:

> has anybody made proposals for talks about PyObjC-related stuff
> at EPC 2003 in Charlreroi? Something about the core pyobjc bridge
> technology and "lessons learned" would be interesting. And maybe
> something about Cocoa itself used from Python would also be nice.

Yes.  Ronald is giving a talk (well, he's submitted one and it has a
good chance <wink> of being accepted) on the bridge.

> I can think about giving an overview talk about Cocoa if nobody
> else wants to. In that case I might be teasing this list for an
> exchange of ideas and input, though... ;-)

I've been emailing with Just about a more user oriented talk, but less
is settled here I think (like for instance which track it belongs in).

Cheers,
M.

-- 
  About the use of language: it is impossible to sharpen a
  pencil with a blunt axe.  It is equally vain to try to do
  it with ten blunt axes instead.
      -- E.W.Dijkstra, 18th June 1975. Perl did not exist at the time.

[Pyobjc-dev] Talk status for EPC 2003?

[Dinu G. <gh...@da...>]

Hi,

has anybody made proposals for talks about PyObjC-related stuff
at EPC 2003 in Charlreroi? Something about the core pyobjc bridge
technology and "lessons learned" would be interesting. And maybe
something about Cocoa itself used from Python would also be nice.

I can think about giving an overview talk about Cocoa if nobody
else wants to. In that case I might be teasing this list for an
exchange of ideas and input, though... ;-)

Regards,

Dinu

--
Dinu C. Gherman
......................................................................
"One of the great things about books is sometimes there are some
fantastic pictures."
(George W. Bush, 3 Jan. 2000)

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Just v. R. <ju...@le...>]

Btw. it's probably better for the list if you send feedback directly to
me. I'll summarize later.

Thanks for the feedback so far! Keep it coming. (Please be clear in your
answers, mention whether you see a solid white rectangle or a striped
rectangle.)

Just

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Dinu G. <gh...@da...>]

I wrote:

>> (Should show a window with a white rectangle, but shows vertical lines
>> of some color(s) on my box that change when resizing the window.)
>
> Does so on my iBook 600 MHz, OS X 10.2.5, /usr/bin/python 2.2.2,
> ATY RageM3p29s (display).

Sorry, that was maybe not so clear. So: it does show a white
rectangle and no other "special" effects!

Dinu

--
Dinu C. Gherman
......................................................................
"If you tell the truth, you don't have to remember anything."
(Mark Twain)

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Dinu G. <gh...@da...>]

Just van Rossum:

>   http://just.letterror.com/~just/DumbApp.app.sit (668 k)
>
> (Should show a window with a white rectangle, but shows vertical lines
> of some color(s) on my box that change when resizing the window.)

Does so on my iBook 600 MHz, OS X 10.2.5, /usr/bin/python 2.2.2,
ATY RageM3p29s (display).

Dinu

--
Dinu C. Gherman
......................................................................
"IBM put the 'blitz' in 'blitzkrieg'. The whole war effort was
organized on Hollerith machines." (Edwin Black)

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Zachery B. <zb...@ur...>]

On Wednesday, Apr 23, 2003, at 14:20 US/Eastern, Just van Rossum wrote:

> (Should show a window with a white rectangle, but shows vertical lines
> of some color(s) on my box that change when resizing the window.)

Don't get that behavior here:

   dual-USB 500MHz iBook
   Card Name: ATY,RageM3p29s
   Card Model: ATY,RageM3
   10.2.5
   /usr/bin/python

Zac

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Just v. R. <ju...@le...>]

bb...@ma... wrote:

> I have tested it on a Dual Proc G4 450 w/a Rage 128 Pro [garbage]
> video card and a TiBook 667 (first TiBook with a decent vid card).
> 
> Both work fine.

Beh. These are the machines I get reliable failure reports from:

    1Ghz Powerbook, ATI Mobility Radeon 9000 AGP
    Card Type:   ATY,DDParent
    Card Name:   ATY,XiaParent
    Card Model:  ATY,RV250M9
    
    400 Mhz TiBook
    Card Type:   RageM3p-parent
    Card Name:   ATY,RageM3p12Parent
    Card Model:  ATY,RageM3
    
    Dual 1.25GHz PowerPC G4 
    ATI Radeon 9000 Pro, 64 Mb. 
    
Just in case it's my *build* setup rather than anything else (eg. I
don't have the latest dev tools installed, dunno exactly which version
actually), here's a standalone app (using Python 2.3 from fairly recent
CVS):

  http://just.letterror.com/~just/DumbApp.app.sit (668 k)

(Should show a window with a white rectangle, but shows vertical lines
of some color(s) on my box that change when resizing the window.)

Thanks for any feedback,

Just

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[<bb...@ma...>]

On Tuesday, Apr 22, 2003, at 16:49 US/Eastern, Just van Rossum wrote:
> I've attached a tiny test program that demonstrates a problem that
> showed up with OS 10.2.5: all of a sudden custom views don't get erased
> properly, but instead filled with a pattern that appears to be random.
> The very same program works well in 10.2.4 and before, regardless of 
> the
> PyObjC version. The equivalent ObjC program works just fine regardless
> of OS version. Does anyone have a clue what could be wrong?

I have tested it on a Dual Proc G4 450 w/a Rage 128 Pro [garbage] video 
card and a TiBook 667 (first TiBook with a decent vid card).

Both work fine.

b.bum

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[<bb...@ma...>]

On Tuesday, Apr 22, 2003, at 17:41 US/Eastern, Just van Rossum wrote:
> Can anyone else see the problem? (I _know_ it's not a problem unique to
> my machine as I got two independent reports from people using two
> different PyObjC apps that see the same as I do on 10.2.5.)
>
> The effect only shows for untransformed rectangles: non-rectangular
> bezier paths don't have the problem.

What video cards do you have in the test systems?

I'll try the app across the other 10.2.5 [6l29] systems I have here.

b.bum

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[Just v. R. <ju...@le...>]

bb...@ma... wrote:

> Working here on 6L28 with /usr/bin/python.

Hm, I get the pattern with both /usr/bin/python and a non-framework
install of CVS Python of a little while back. This is on 6L29. PyObjC
straight from CVS, installed with distutils.

Can anyone else see the problem? (I _know_ it's not a problem unique to
my machine as I got two independent reports from people using two
different PyObjC apps that see the same as I do on 10.2.5.)

The effect only shows for untransformed rectangles: non-rectangular
bezier paths don't have the problem.

Just

Re: [Pyobjc-dev] scary problem with custom view & 10.2.5

[<bb...@ma...>]

Working here on 6L28 with /usr/bin/python.

On Tuesday, Apr 22, 2003, at 16:49 US/Eastern, Just van Rossum wrote:
> I've attached a tiny test program that demonstrates a problem that
> showed up with OS 10.2.5: all of a sudden custom views don't get erased
> properly, but instead filled with a pattern that appears to be random.
> The very same program works well in 10.2.4 and before, regardless of 
> the
> PyObjC version. The equivalent ObjC program works just fine regardless
> of OS version. Does anyone have a clue what could be wrong?

[Pyobjc-dev] scary problem with custom view & 10.2.5

[Just v. R. <ju...@le...>]

I've attached a tiny test program that demonstrates a problem that
showed up with OS 10.2.5: all of a sudden custom views don't get erased
properly, but instead filled with a pattern that appears to be random.
The very same program works well in 10.2.4 and before, regardless of the
PyObjC version. The equivalent ObjC program works just fine regardless
of OS version. Does anyone have a clue what could be wrong?

Just

Re: [Pyobjc-dev] Nuts! Project Templates Broken!

[<bb...@ma...>]

On Saturday, Apr 19, 2003, at 13:25 US/Eastern, Bill Bumgarner wrote:
> The templates are broken if installed via the package;  fixing now.  
> Sorry!

The specifications and templates have been moved back to /Developer due 
to a bug in PBX.

Specifically, PBX looks in /Network/Developer, /Developer, and 
~/Developer for the additional resources-- but fails to look in 
/Library/Developer (as it should).

So, for now, the PB Extras are in /Developer again.   That is truly 
unfortunate in that it means the developer will have to reinstall 
PyObjC after removing/upgrading the dev tools.

A bug has been filed at bugreport.apple.com.

*sigh*

b.bum

[Pyobjc-dev] Nuts! Project Templates Broken!

[Bill B. <bb...@co...>]

The templates are broken if installed via the package;  fixing now.  
Sorry!

[Pyobjc-dev] 0.9 packaging fixed

[Bill B. <bb...@co...>]

I completed the changes to the 0.9 packaging and have restructured 
everything related to ProjectBuilder into the ProjectBuilder Extras 
directory within the repository.    I also added the Python language 
definition for Project Builder;  PB will now colorize Python code once 
you install 0.9.

Everything not in site-packages is now installed-- documentation, 
examples, templates, etc..-- into /Library/Developer/.

When installing over 0.9pre1, Installer should correctly upgrade the 
system by removing the files from their original location.  
post_install and post_upgrade were conslidated into preflight -- 
preflight now also removes the 0.8 examples, templates, and receipt.

b.bum

Begin forwarded commit message:
> Log Message:
> Consolidated post_install and post_upgrade into preflight script.  It 
> is safer to delete legacy cruft prior to installing new cruft.  
> Integrated removal of examples and templates from the pre-0.9 location 
> of /Developer.
>
> Finished moving templates and language specifications to 
> /Library/Developer as per Apple's informal/unsupported recommendation. 
>  Documentation and Examples can also be found in /Library/Developer 
> now.

[Pyobjc-dev] Upgrade package

[Bill B. <bb...@co...>]

It appears that we cannot upgrade from PyObjC 0.8.0.pkg unless we call 
the 0.9 package the same thing.

Bummer.

As such, we probably need a pre-flight that detects the presence of the 
PyObjC 0.8.0 (or 0.7.0) installer package receipt and then uninstalls 
those packages prior to installing PyObjC.pkg.

Now that the package is generically named, it'll "just work" going 
forward (I just tested it).

I'm working on this problem now.

b.bum