rubycocoa-talk — talking about RubyCocoa

to close ;-) Re: [Rubycocoa-talk] standaloneify.rb

[Yvon T. <tho...@fr...>]

Le 23 f=E9vr. 06 =E0 10:23, Jonathan Paisley a =E9crit :

> Hi,
>
> I looked at the package which seems to be working properly, with =20
> dependencies. The ruby runtime is linked into the RubyCocoa =20
> framework so it should be completely independent of any ruby on the =20=

> host system.
>
>> he is using this RubyCocoa beta app over standard Apple Ruby, i'm =20
>> using the latest build by my own.
>
> Could you explain what you mean here a bit more? Is he attempting =20
> to run the same version you provided a link to? If he's using a =20
> standalone version that you've provided, even though he has only =20
> the standard Apple ruby, it will be using the latest build that you =20=

> used to make the application.
>
>
> For more debugging information, you could add 'p $LOAD_PATH' to =20
> rb_main.rb before the first 'require'. Another useful thing is to =20
> run the app from the terminal using 'ktrace':
>
>  ktrace ./BookmarksMerge.app/Contents/MacOS/BookmarksMerge
>  kdump | grep NAMI > files.log
>
> 'files.log' will show any file accesses made by the application, =20
> which should help to track down where it's looking for 'nkf.bundle'.
>


on one computer a "G5 mono 1.8" it seems the prob comes from the =20
computer itself see  "The Buggiest Macintosh. Ever?(TM)"
<http://www.g5freeze.com/index.html>

on the other the bugg gad nothing to do with nkf nor standalonify.rb.


thanks for your help !

Yvon=

Re: [Rubycocoa-talk] Bluetooth

[Jonathan P. <jp...@dc...>]

> hmm, as I never code in objective-c before, I don't
> know exactly what NSApplication.sharedApplication.run
> should do, but I'll read the documentation and
> feedback you.

According to this page:

   http://www.cocoadev.com/index.pl?IOBluetooth

the bluetooth APIs are not well documented. I think it may be best to  
try to make an objective C sample program that works before trying it  
in RubyCocoa, just to eliminate any interference from the RubyCocoa  
bridge.

The best place to get advice is probably one of the Cocoa development  
lists, rather than ruby-related lists.

Good luck,
Jonathan

Re: [Rubycocoa-talk] Bluetooth

[Davide V. <var...@ya...>]

Hi Jonathan

--- Jonathan Paisley <jp...@dc...> wrote:

> On 23 Feb 2006, at 22:35, Davide Varvello wrote:
> 
> > I know the channel id for ftp on my device is 9,
> puts
> > connectResult
> > displays 0, but puts sendResult shows -21875 (i.e
> > busy) and of course,
> > no picture is sent to the device.
> > Do you understand this behavior?
> 
> That error number is kOBEXSessionBusyError (see
> /System/Library/ 
> Frameworks/IOBluetooth.framework/Headers/OBEX.h).

yep

> I think the problem is that the the OBEX methods are
> asynchronous -  
> you have to set them going then wait for the
> callback method on the  
> delegate. See the documentation for
> [OBEXFileTransferServices  
> setDelegate:].

right

> The reason progress is not made even though you have
> 'sleep' in your  
> code is that the Cocoa event loop doesn't get a
> chance to run. Try  
> setting up a delegate object and initiating the
> session, then kick  
> off the event loop with
> NSApplication.sharedApplication.run.


hmm, as I never code in objective-c before, I don't
know exactly what NSApplication.sharedApplication.run
should do, but I'll read the documentation and
feedback you.
Thanks
 Davide


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: [Rubycocoa-talk] rubycocoa game example

[Tim B. <tim...@gm...>]

Jonathan,

yes, that suggestion sounds great, especially since it would make  
things symmetric.

Tim

On Feb 23, 2006, at 12:48 AM, Jonathan Paisley wrote:

>
> On 22 Feb 2006, at 23:56, Tim Burks wrote:
>
>> I'd prefer to have the bridge automatically convert "object=" into  
>> "setObject" calls, perhaps as an option.
>
> There are two directions to consider here:
>
>   ObjC calling to Ruby for 'setFoo' to call 'foo='. This is handled  
> by kvc_accessor.
>
>   Ruby calling 'foo=' on an ObjC object to call 'setFoo'. This  
> isn't handled yet.
>
> The latter could be fixed by extending the 'analyze_missing' method  
> in oc_wrapper.rb. Do you think that would be sufficient?
>
>

Re: [Rubycocoa-talk] Bluetooth

[Jonathan P. <jp...@dc...>]

On 23 Feb 2006, at 22:35, Davide Varvello wrote:

> I know the channel id for ftp on my device is 9, puts
> connectResult
> displays 0, but puts sendResult shows -21875 (i.e
> busy) and of course,
> no picture is sent to the device.
> Do you understand this behavior?

That error number is kOBEXSessionBusyError (see /System/Library/ 
Frameworks/IOBluetooth.framework/Headers/OBEX.h).

I think the problem is that the the OBEX methods are asynchronous -  
you have to set them going then wait for the callback method on the  
delegate. See the documentation for [OBEXFileTransferServices  
setDelegate:].

The reason progress is not made even though you have 'sleep' in your  
code is that the Cocoa event loop doesn't get a chance to run. Try  
setting up a delegate object and initiating the session, then kick  
off the event loop with NSApplication.sharedApplication.run.

[Rubycocoa-talk] Bluetooth

[Davide V. <var...@ya...>]

Hi,
 I've already post this message to the ruby group, but
no one replied me, I hope for your help. So...

I'm trying to connect, over bluetooth, to a mobile
device and send apicure.
That's my code:

require 'osx/cocoa'

OSX::NSBundle.bundleWithPath("/System/Library/Frameworks/IOBluetoothUI.framework").load

OSX.ns_import :IOBluetoothDeviceSelectorController

bluedevices =
OSX::IOBluetoothDeviceSelectorController.deviceSelector
bluedevices.runModal;
result = bluedevices.getResults;

bluetoothDevice = result.objectAtIndex(0);

OSX::NSBundle.bundleWithPath("/System/Library/Frameworks/IOBluetooth.framework").load;
OSX.ns_import :IOBluetoothOBEXSession;
obexSession =
OSX::IOBluetoothOBEXSession.withDevice_channelID(bluetoothDevice,
9);

OSX.ns_import :OBEXFileTransferServices;
service =
OSX::OBEXFileTransferServices.withOBEXSession(obexSession);

connectResult = service.connectToFTPService;
puts connectResult;
sleep 3;
sendResult=service.sendFile("/Users/davide/myPicture.jpeg");
puts sendResult;
sleep 3;
service.disconnect;

I know the channel id for ftp on my device is 9, puts
connectResult
displays 0, but puts sendResult shows -21875 (i.e
busy) and of course,
no picture is sent to the device.
Do you understand this behavior?
TIA
 Davide 

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Re: [Rubycocoa-talk] standaloneify.rb

[Yvon T. <tho...@fr...>]

Le 23 f=E9vr. 06 =E0 10:23, Jonathan Paisley a =E9crit :

> I looked at the package which seems to be working properly, with =20
> dependencies. The ruby runtime is linked into the RubyCocoa =20
> framework so it should be completely independent of any ruby on the =20=

> host system.

it's what i was assuming )))

>
>> he is using this RubyCocoa beta app over standard Apple Ruby, i'm =20
>> using the latest build by my own.
>
> Could you explain what you mean here a bit more? Is he attempting =20
> to run the same version you provided a link to? If he's using a =20
> standalone version that you've provided, even though he has only =20
> the standard Apple ruby, it will be using the latest build that you =20=

> used to make the application.

yes he used the standalone version, he had been using another app =20
(also standalonified by your script) with success.

>
>
> For more debugging information, you could add 'p $LOAD_PATH' to =20
> rb_main.rb before the first 'require'. Another useful thing is to =20
> run the app from the terminal using 'ktrace':
>

ok, i see what you mean, before using your script, i've been using =20
one of mine, but really less efficient (the app got around 14 MO, =20
with your script saying 4 MO !)

>  ktrace ./BookmarksMerge.app/Contents/MacOS/BookmarksMerge
>  kdump | grep NAMI > files.log
>
> 'files.log' will show any file accesses made by the application, =20
> which should help to track down where it's looking for 'nkf.bundle'.

ok, fine, i'll ask him to do that.

it seems the prob is computer dependant because on one computer he =20
had the message previously given and on another the application =20
starts up attempts to convert safari and|or firebox bookmarks to xml =20
and crash when sending the results to NSOutlineView.

the latest bugg is more clearer to me (the safari and firefox =20
bookmarks files aren't very well structured then...)

best, and thanks a lot !

Yvon

Re: [Rubycocoa-talk] standaloneify.rb

[Jonathan P. <jp...@dc...>]

Hi,

I looked at the package which seems to be working properly, with  
dependencies. The ruby runtime is linked into the RubyCocoa framework  
so it should be completely independent of any ruby on the host system.

> he is using this RubyCocoa beta app over standard Apple Ruby, i'm  
> using the latest build by my own.

Could you explain what you mean here a bit more? Is he attempting to  
run the same version you provided a link to? If he's using a  
standalone version that you've provided, even though he has only the  
standard Apple ruby, it will be using the latest build that you used  
to make the application.


For more debugging information, you could add 'p $LOAD_PATH' to  
rb_main.rb before the first 'require'. Another useful thing is to run  
the app from the terminal using 'ktrace':

  ktrace ./BookmarksMerge.app/Contents/MacOS/BookmarksMerge
  kdump | grep NAMI > files.log

'files.log' will show any file accesses made by the application,  
which should help to track down where it's looking for 'nkf.bundle'.

Re: [Rubycocoa-talk] rubycocoa game example

[Jonathan P. <jp...@dc...>]

On 22 Feb 2006, at 23:56, Tim Burks wrote:

> I'd prefer to have the bridge automatically convert "object=" into  
> "setObject" calls, perhaps as an option.

There are two directions to consider here:

   ObjC calling to Ruby for 'setFoo' to call 'foo='. This is handled  
by kvc_accessor.

   Ruby calling 'foo=' on an ObjC object to call 'setFoo'. This isn't  
handled yet.

The latter could be fixed by extending the 'analyze_missing' method  
in oc_wrapper.rb. Do you think that would be sufficient?

Re: [Rubycocoa-talk] standaloneify.rb

[Yvon T. <tho...@fr...>]

Le 23 f=E9vr. 06 =E0 00:19, Jonathan Paisley a =E9crit :

> It's strange that 'nkf' fails to load even though it's been =20
> included. Probably a bug in standaloneify. Can you send me a zip of =20=

> the original and 'standaloneified' app to investigate?
>
>


here it is :

> <http://81.57.96.20/BookmarksMerge-200602221356.dmg.bz2>

beware that' supposed to read Safari bookmarks and firefox too, if =20
not would crash after loading for other reasons, but works on my =20
computer.

i could package the latest if you want.

best,

Yvon=

Re: [Rubycocoa-talk] rubycocoa game example

[Tim B. <tim...@gm...>]

On Feb 22, 2006, at 12:41 PM, Jonathan Paisley wrote:

> On 22 Feb 2006, at 18:16, Tim Burks wrote:
>
>> * Something needs to be done to resolve the different calling  
>> conventions used by Ruby attribute setters ("object =") and Cocoa  
>> KVC setters ("setObject").
>
> Have you any ideas for how best to resolve this?

I used the hack below to replace "attr_accessor" with "cocoa_accessor".

cocoa_accessor creates a "setObject" method instead of "object=".

I'd prefer to have the bridge automatically convert "object=" into  
"setObject" calls, perhaps as an option.

Tim

module CocoaAccessor
     module ClassMethods
         def cocoa_accessor(value_id)
             iv_name = "@#{value_id}".to_sym
             define_method(value_id) { self.instance_variable_get 
(iv_name ) }
             define_method("set#{value_id.to_s.capitalize}".to_sym) {| 
x| self.instance_variable_set(iv_name, x) }
         end
     end
     def self.append_features(klass)
         super
         klass.extend(ClassMethods)
     end
end

class RubySprite
     include CocoaAccessor
     cocoa_accessor :position
     cocoa_accessor :direction
     cocoa_accessor :velocity
     cocoa_accessor :radius
     cocoa_accessor :color

Re: [Rubycocoa-talk] standaloneify.rb

[Jonathan P. <jp...@dc...>]

On 22 Feb 2006, at 18:32, Yvon Thoraval wrote:

>> /Applications/BookmarksMerge.app/Contents/Frameworks/ 
>> RubyCocoa.framework/
>> Versions/A/Resources/ruby/osx/objc/cocoa_macros.rb:13:in  
>> `require': no
>> such file to load -- nkf (LoadError)
>> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/ 
>> cocoa_macros.rb:13
>> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/cocoa.rb:13
>> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/cocoa.rb:11
>> 	from /Applications/BookmarksMerge.app/Contents/Resources/ 
>> rb_main.rb:25
>

> i've carefully inspected the Contents folder of each app and catch  
> for any difference : found no differences at all ( apart from what  
> is specific to the app).
>
> even i've found the "nkf.bundle" :

> -rwx------    1 yvon  yvon  222304 Feb 22 13:56 nkf.bundle

It's strange that 'nkf' fails to load even though it's been included.  
Probably a bug in standaloneify. Can you send me a zip of the  
original and 'standaloneified' app to investigate?

Re: [Rubycocoa-talk] standaloneify.rb

[Neil S. <ne...@ha...>]

Yvon Thoraval wrote:
> Hey all,
>=20
> i'm using "standaloneify.rb" from Jonathan Paisley
>=20
> and, for the last app i'm working on, a user said to the app crashes
> with the following message at Cosole log :
>=20
>=20
>> /Applications/BookmarksMerge.app/Contents/Frameworks/RubyCocoa.framewo=
rk/
>> Versions/A/Resources/ruby/osx/objc/cocoa_macros.rb:13:in `require': no=

>> such file to load -- nkf (LoadError)
>>     from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/cocoa_macros.rb=
:13
>>     from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/cocoa.rb:13
>>     from /Applications/BookmarksMerge.app/Contents/Frameworks/
>> RubyCocoa.framework/Versions/A/Resources/ruby/osx/cocoa.rb:11
>>     from
>> /Applications/BookmarksMerge.app/Contents/Resources/rb_main.rb:25

If you run into something that standaloneify doesn't pick up on its own,
because of runtime-added requires or whatnot, use the -l option to add th=
em.

So here you want to add -l nkf to your standaloneify call and you should
be fine.


--=20
Neil Stevens - ne...@ha...
"There is nothing patriotic about hating your country, or pretending
that you can love your country but despise your government."
 -- William Jefferson Clinton

Re: [Rubycocoa-talk] rubycocoa game example

[Jonathan P. <jp...@dc...>]

On 22 Feb 2006, at 18:16, Tim Burks wrote:

> One possible optimization is to replace the Sprite class with a  
> version written in Objective-C.   Ideally, we could just swap in a  
> new parent class with no changes to its subclasses.  But in  
> practice, I had several problems:
>
> * The Ruby versions of Sprite and its subclasses must already be  
> using Objective-C alloc/init construction instead of new.
>
> * Methods in Ruby subclasses that will be replaced by Objective-C  
> classes must map to Objective-C message names.

This is true - but in RubyCocoa you never need the trailing  
underscore in method names. You do need it, however, when calling  
ns_overrides.

> * Something needs to be done to resolve the different calling  
> conventions used by Ruby attribute setters ("object =") and Cocoa  
> KVC setters ("setObject").

Have you any ideas for how best to resolve this?

> * When a Ruby subclass overrides an Objective-C class method, an  
> "ns_overrides" declaration seems to be necessary.  But if the  
> parent class is switched back to Ruby (for debugging or testing),  
> that "ns_overrides" is a fatal error.

Good point. I'm hoping it might be possible to do away with  
ns_overrides altogether...

> Surprisingly, I found that replacing the Ruby Sprite class with an  
> Objective-C version didn't speed up my game; in fact it seemed to  
> slow it down a bit.  The reason for this was that the new version  
> was making more calls across the bridge, and calls across the  
> bridge are much more expensive than code implemented in Ruby or  
> calls to standard Ruby C extensions.  I made some crude  
> measurements and estimated that the best I could hope for was a few  
> thousand bridge crossings per second.  That's plenty for a typical  
> Cocoa app, but not enough for an arcade game that uses RubyCocoa to  
> draw and manage objects.  So I switched back to my Ruby Sprite  
> class and rewrote my object drawing functions as standard Ruby C  
> extensions.  After that I had only 2-3 bridge crossings per frame  
> and the game was snappy again, even with the brute-force Ruby  
> collision detection that I had worried about initially.

I think the drawing is probably one of the slowest things. Have you  
tried getting a profile using Shark? I looked briefly and noticed  
that NSWindow clears the window as well as the draw method in Game.  
Together these took up about 20% of CPU time. Turning off anti- 
aliasing would speed things up too. Or using OpenGL! :)

I would hope that using plain ruby Sprite objects and trying to keep  
as much as possible in Ruby-land (e.g., only calling ObjC for the  
drawing) should give reasonable performance. I agree that bridge  
crossings are expensive.

Cheers,
Jonathan

[Rubycocoa-talk] standaloneify.rb

[Yvon T. <tho...@fr...>]

Hey all,

i'm using "standaloneify.rb" from Jonathan Paisley

and, for the last app i'm working on, a user said to the app crashes  
with the following message at Cosole log :


> /Applications/BookmarksMerge.app/Contents/Frameworks/ 
> RubyCocoa.framework/
> Versions/A/Resources/ruby/osx/objc/cocoa_macros.rb:13:in `require': no
> such file to load -- nkf (LoadError)
> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/ 
> cocoa_macros.rb:13
> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
> RubyCocoa.framework/Versions/A/Resources/ruby/osx/objc/cocoa.rb:13
> 	from /Applications/BookmarksMerge.app/Contents/Frameworks/
> RubyCocoa.framework/Versions/A/Resources/ruby/osx/cocoa.rb:11
> 	from /Applications/BookmarksMerge.app/Contents/Resources/ 
> rb_main.rb:25

he is using this RubyCocoa beta app over standard Apple Ruby, i'm  
using the latest build by my own.

the dependencies are :

require 'osx/cocoa'
require 'osx/aeosa'
require 'iconv'
require 'date'
require 'yaml'
require 'xml/xslt'
require 'rexml/document'

written directly into "rb_main.rb"

it's surprising to me because i've used "standaloneify.rb" for  
another RubyCocoa app having the same dependencies as listed above.

i've carefully inspected the Contents folder of each app and catch  
for any difference : found no differences at all ( apart from what is  
specific to the app).

even i've found the "nkf.bundle" :

./lib:
total 2592
drwxr-xr-x   10 yvon  yvon     340 Feb 22 13:56 .
drwxr-xr-x    9 yvon  yvon     306 Feb 22 13:56 ..
-rwx------    1 yvon  yvon    9992 Feb 22 13:56 fcntl.bundle
-rwx------    1 yvon  yvon   33204 Feb 22 13:56 iconv.bundle
-rwx------    1 yvon  yvon  222304 Feb 22 13:56 nkf.bundle
-rwx------    1 yvon  yvon  681728 Feb 22 13:56 openssl.bundle
-rwx------    1 yvon  yvon   50484 Feb 22 13:56 osx_aeosa.bundle
-rwx------    1 yvon  yvon   37904 Feb 22 13:56 stringio.bundle
-rwx------    1 yvon  yvon  272464 Feb 22 13:56 syck.bundle
drwxr-xr-x    3 yvon  yvon     102 Feb 22 13:56 xml

./lib/xml:
total 152
drwxr-xr-x    3 yvon  yvon    102 Feb 22 13:56 .
drwxr-xr-x   10 yvon  yvon    340 Feb 22 13:56 ..
-rwx------    1 yvon  yvon  76972 Feb 22 13:56 xslt.bundle


same perms, same sizes in each case.

The only major difference in my code it's in that i'm using a proxy  
class (advice from onathan Paisley).

what could i ask to the user in order to go ahead ?


obviously, my app is working as standalone on my computer.

here is my ruby setup :
 > ruby --version
ruby 1.8.4 (2005-12-24) [powerpc-darwin8.4.0]

being in :
 > which ruby
/usr/local/bin/ruby


best,

Yvon

Re: [Rubycocoa-talk] rubycocoa game example

[Jonathan P. <jp...@dc...>]

On 22 Feb 2006, at 18:16, Tim Burks wrote:

> p.s. You suggested the following procedure for setting the Document  
> class:
>
> > Rather than editing the pbxproj file manually:
> >   Targets inspector -> rubyrocks -> Info.plist Entries ->  
> Document Types -> Document Class
>
> Have you verified this? If you can do this for the rubyrocks  
> project, would you please send me a screenshot? I may be missing  
> something obvious; it's not working for me.  I know that it's  
> possible for Cocoa Document-based apps, but the RubyCocoa project  
> template seems to be different.  When I look at the rubyrocks  
> target with the inspector, I only see two tabs: "General" and  
> "Comments".  I don't see how I could use either to modify the  
> Info.plist file.  I'm using Xcode 2.2.1.  Thanks!

I used the method while working through the beginning of the  
tutorial, since I refused to edit the project file :)

I've included a screenshot of the target inspector I have below.  
Don't inspect the  target - double click on it (I've no idea how to  
get to this otherwise).

The RubyCocoa template projects are old-style Xcode projects. If you  
do 'Project -> Upgrade to Native Target' you can then edit the  
document class in the Properties tab of the inspector.

Cheers,
Jonathan

Re: [Rubycocoa-talk] rubycocoa game example

[Tim B. <tim...@gm...>]

Jonathan,

Thanks for your comments.  My example was extracted from a more  
complex game project that I've been using to investigate RubyCocoa.   
I like using the document-based architecture because it lets me log  
game events and easily save and restore them for game replays, which  
are great for demos, debugging, and performance analysis.

My "advanced" version of the game supports two players and subdivides  
rocks (large rocks break into two medium-sized rocks, medium-sized  
rocks break into two small rocks, which then can be destroyed).  The  
result is that there are a lot more objects flying around, which  
starts to slow down the game (even when I disable collision  
detection, which is an obvious potential bottleneck).

One possible optimization is to replace the Sprite class with a  
version written in Objective-C.   Ideally, we could just swap in a  
new parent class with no changes to its subclasses.  But in practice,  
I had several problems:

* The Ruby versions of Sprite and its subclasses must already be  
using Objective-C alloc/init construction instead of new.

* Methods in Ruby subclasses that will be replaced by Objective-C  
classes must map to Objective-C message names.

* Something needs to be done to resolve the different calling  
conventions used by Ruby attribute setters ("object =") and Cocoa KVC  
setters ("setObject").

* When a Ruby subclass overrides an Objective-C class method, an  
"ns_overrides" declaration seems to be necessary.  But if the parent  
class is switched back to Ruby (for debugging or testing), that  
"ns_overrides" is a fatal error.

Surprisingly, I found that replacing the Ruby Sprite class with an  
Objective-C version didn't speed up my game; in fact it seemed to  
slow it down a bit.  The reason for this was that the new version was  
making more calls across the bridge, and calls across the bridge are  
much more expensive than code implemented in Ruby or calls to  
standard Ruby C extensions.  I made some crude measurements and  
estimated that the best I could hope for was a few thousand bridge  
crossings per second.  That's plenty for a typical Cocoa app, but not  
enough for an arcade game that uses RubyCocoa to draw and manage  
objects.  So I switched back to my Ruby Sprite class and rewrote my  
object drawing functions as standard Ruby C extensions.  After that I  
had only 2-3 bridge crossings per frame and the game was snappy  
again, even with the brute-force Ruby collision detection that I had  
worried about initially.

So to conclude, I found RubyCocoa to be great for prototyping,  
designing, and testing.  I used code that I had written in Ruby to  
guide C and Objective-C rewrites.   If I take this project further, I  
expect that I will rewrite more of the game objects and their  
management in Objective-C and use Ruby for testing, top-level game  
control, and user input handling.

Tim

p.s. You suggested the following procedure for setting the Document  
class:

 > Rather than editing the pbxproj file manually:
 >   Targets inspector -> rubyrocks -> Info.plist Entries -> Document  
Types -> Document Class

Have you verified this? If you can do this for the rubyrocks project,  
would you please send me a screenshot? I may be missing something  
obvious; it's not working for me.  I know that it's possible for  
Cocoa Document-based apps, but the RubyCocoa project template seems  
to be different.  When I look at the rubyrocks target with the  
inspector, I only see two tabs: "General" and "Comments".  I don't  
see how I could use either to modify the Info.plist file.  I'm using  
Xcode 2.2.1.  Thanks!

On Feb 22, 2006, at 2:36 AM, Jonathan Paisley wrote:

>
> On 22 Feb 2006, at 7:23, Tim Burks wrote:
>
>> This evening I spoke to the San Francisco Ruby Meetup group about  
>> RubyCocoa.  I thought it would be fun to describe RubyCocoa with  
>> an example, so I presented a simple arcade-style game written in  
>> Ruby that uses RubyCocoa to draw and handle user events.  I also  
>> wrote it up as a tutorial and put it online here: http:// 
>> www.rubycocoa.com/ruby-rocks.
>
> Great stuff!
>
> A few comments on the tutorial:
>
> Probably worth mentioning up front that a game wouldn't normally be  
> a document-based application (and say at the end what would be  
> different to use a non-document application template).
>
> Rather than editing the pbxproj file manually:
>   Targets inspector -> rubyrocks -> Info.plist Entries -> Document  
> Types -> Document Class
>
> I couldn't see any instructions for binding GameView.game to the  
> Game instance (File's Owner) in Interface Builder.
>
> The instructions for adding the 'game' outlet aren't as detailed as  
> the result of the instructions in Interface Builder.
>
> You can just use [a,b,c,d] instead of OSX::NSRect.new(a,b,c,d)  
> (similarly for NSPoint), unless you want to hang on to the object  
> and use .x, .y etc.
>
> The moveWithBounds_ and initWithPosition_ methods etc don't need a  
> trailing underscore.
>
> The model objects (Sprite and descendents) can be plain Ruby  
> objects rather than NSObject descendents. This means you can use a  
> normal ruby initialize method (instead of .alloc.initWith...).
>
> Cheers,
> Jonathan
>
>
>

Re: [Rubycocoa-talk] rubycocoa game example

[Tim B. <tim...@gm...>]

Hi Phil,

I'm using a slightly customized version of Tobias L=FCtke's hieraki-0.5.
(hieraki-0.5 is available here: http://dist.leetsoft.com/pkg)

I added a sitemap and permalinks but haven't released my changes, =20
because the hieraki project has gone through some significant =20
revisions since 0.5.  More recent versions contain a lot of =20
improvements for authors, but I felt that 0.5's simpler interface was =20=

more intuitive for my readers.

Tim

On Feb 22, 2006, at 12:24 AM, Phil Tomson wrote:

>
>
> On Tue, 21 Feb 2006, Tim Burks wrote:
>
>>
>> This evening I spoke to the San Francisco Ruby Meetup group about
>> RubyCocoa.  I thought it would be fun to describe RubyCocoa with an
>> example, so I presented a simple arcade-style game written in Ruby
>> that uses RubyCocoa to draw and handle user events.  I also wrote it
>> up as a tutorial and put it online here: http://www.rubycocoa.com/
>> ruby-rocks.
>>
>> Tim
>>
>
> This is cool!  thanks for putting it together.
>
> BTW: how did you generate the webpages?  Did you use hieraki?
>
> Phil
>

Re: [Rubycocoa-talk] Array of NSButton ???

[Yvon T. <tho...@fr...>]

Le 22 f=E9vr. 06 =E0 14:02, kimura wataru a =E9crit :

> I think it is an easy way to use thread.

i've tried that, however this doesn't correspond to my needs, =20
because, with thread, we might launch asynchronously any action from =20
each other, which is not what i want.

best,

Yvon=

Re: [Rubycocoa-talk] Array of NSButton ???

[Yvon T. <tho...@fr...>]

Le 22 f=E9vr. 06 =E0 13:52, Jonathan Paisley a =E9crit :

>
> Hehe - it's calling the ruby method 'display'. Try 'oc_display' =20
> instead. This bypasses any ruby methods defined.

#oc_display works fine.

however, something surprising to me is when a button is disabled, it =20
still memorizes click on it, only the view is changing.

that's Cocoa design :[

(i never had such behaviour with Java-Swing...)

>
>> then i went to "discardEventsMatchingMask:beforeEvent:" and =20
>> applied to a NSWindow as a receiver :
>>
>> @mainWindow.discardEventsMatchingMask_beforeEvent=20
>> (OSX::NSAnyEventMask, OSX::NSKeyUpMask)
>
> Ah, I realise now that discartEvents... won't work here.
>
> Instead look at "- (NSEvent *)nextEventMatchingMask:(unsigned int)=20
> mask untilDate:(NSDate *)expiration inMode:(NSString *)mode dequeue:=20=

> (BOOL)flag" and call it repeatedly with dequeue:YES and try =20
> different values for untilDate: NSDate.date, NSDate.distantFuture =20
> or NSData.distantPast. Stop calling it when it returns no events.
>

ok, i need to investigate further on that point.

Yvon

Re: [Rubycocoa-talk] Array of NSButton ???

[kimura wataru<ki...@us...>]

Hi,

I think it is an easy way to use thread. But we have a bug report 
ruby's thread on RubyCocoa sometimes crashes. 

  def loadDroite(sender)
    disableAllButtons                     ####### here i disable all buttons
    Thread.new {
      loadXmlDroite("firefox.xml", "Barre personnelle")
      @xmlDroite="firefox.xml"
      @labelDroite.setStringValue "Firefox"
      enableAllButtons                      ####### here i re-enable all buttons
    }
  end

Wed, Feb 22, 2006 11:58:41 AM, Yvon Thoraval wrote:
>	def loadDroite(sender)
>     disableAllButtons                     ####### here i disable all  
>buttons
>     loadXmlDroite("firefox.xml", "Barre personnelle")
>     @xmlDroite="firefox.xml"
>     @labelDroite.setStringValue "Firefox"
>     enableAllButtons                      ####### here i re-enable  
>all buttons
>   end
>
>i do the "same" (isableAllButtons [...] enableAllButtons) for all the  
>actions corresponding to all the buttons.
>


-- 
kimura wataru

Re: [Rubycocoa-talk] Array of NSButton ???

[Jonathan P. <jp...@dc...>]

> with #display i get only that in the XCode's Run log :
>
> [Session started at 2006-02-22 13:13:49 +0100.]
> <NSButton: 0x69d0f0><NSButton: 0x4b3210><NSButton:  
> 0x4bba30><NSButton: 0x6dfc70><NSButton: 0x689400><NSButton:  
> 0x4b47a0><NSButton: 0x6899e0><NSButton: 0x6e5080>
> BookmarksMerge has exited with status 0.

Hehe - it's calling the ruby method 'display'. Try 'oc_display'  
instead. This bypasses any ruby methods defined.

> then i went to "discardEventsMatchingMask:beforeEvent:" and applied  
> to a NSWindow as a receiver :
>
> @mainWindow.discardEventsMatchingMask_beforeEvent 
> (OSX::NSAnyEventMask, OSX::NSKeyUpMask)

Ah, I realise now that discartEvents... won't work here.

Instead look at "- (NSEvent *)nextEventMatchingMask:(unsigned int) 
mask untilDate:(NSDate *)expiration inMode:(NSString *)mode dequeue: 
(BOOL)flag" and call it repeatedly with dequeue:YES and try different  
values for untilDate: NSDate.date, NSDate.distantFuture or  
NSData.distantPast. Stop calling it when it returns no events.

Re: [Rubycocoa-talk] Array of NSButton ???

[Yvon T. <tho...@fr...>]

Le 22 f=E9vr. 06 =E0 12:17, Jonathan Paisley a =E9crit :

> I think the problem is that the visible 'disabled' state only =20
> updates when the event loop runs again. During your long-running =20
> method, the Cocoa event loop never gets a chance to run. Therefore, =20=

> any clicks are queued and get processed later -- by which point the =20=

> buttons have been enabled again.
>

ok, i understand.
> You could try disabling the buttons, explicitly redrawing them by =20
> calling 'display'. You could clear any button clicks from the event =20=

> queue with 'discardEventsMatchingMask:beforeEvent:'.
>

display (eventually after a "setNeedsDisplay true" ) doesn't change =20
anything :

   def disableAllButtons
     @buttons.each { |b|
       b.setEnabled(false)
       #b.setNeedsDisplay true
       b.display
     }
     sleep 1
   end

sleep 1 doesn't help too

with #display i get only that in the XCode's Run log :

[Session started at 2006-02-22 13:13:49 +0100.]
<NSButton: 0x69d0f0><NSButton: 0x4b3210><NSButton: =20
0x4bba30><NSButton: 0x6dfc70><NSButton: 0x689400><NSButton: =20
0x4b47a0><NSButton: 0x6899e0><NSButton: 0x6e5080>
BookmarksMerge has exited with status 0.


may be one drastic solution coud be using #setHidden true gaves no =20
change too...


then i went to "discardEventsMatchingMask:beforeEvent:" and applied =20
to a NSWindow as a receiver :

@mainWindow.discardEventsMatchingMask_beforeEvent=20
(OSX::NSAnyEventMask, OSX::NSKeyUpMask)

here i got this error in the XCode's Run log :

[Session started at 2006-02-22 13:13:18 +0100.]
2006-02-22 13:13:24.417 BookmarksMerge[22686] *** -[NSDecimalNumber =20
_cgsEventTime]: selector not recognized [self =3D 0x4bb0af0]
/Library/Frameworks/RubyCocoa.framework/Versions/A/Resources/ruby/osx/=20=

objc/oc_wrapper.rb:17:in `NSApplicationMain': NSApplicationMain - =20
NSInvalidArgumentException - *** -[NSDecimalNumber _cgsEventTime]: =20
selector not recognized [self =3D 0x4bb0af0] (OSX::OCException)
	from =
/Users/yvon/work/RubyCocoa/BookmarksMerge/build/Development/=20
BookmarksMerge.app/Contents/Resources/rb_main.rb:46

BookmarksMerge has exited with status 1.


i've followed what's written in the Cocoa browser...

?

Yvon

Re: [Rubycocoa-talk] Array of NSButton ???

[Jonathan P. <jp...@dc...>]

On 22 Feb 2006, at 10:58, Yvon Thoraval wrote:

> the appearance of the buttons doesn't change and also if, when an  
> action isn't terminated (that's to say between disableAllButtons  
> and enableAllButtons) the click is memorized and, when the action  
> is finished the process enters a new action corresponding to the  
> new button pressed.
>
> if, in between disableAllButtons and enableAllButtons, i press two  
> times a button and another one afterwards, all the click on all the  
> different buttons and the corresponding actions are undertaken each  
> after each...
>
> i'd like to avoid that : no memorization when an action is undertaken.
>

I think the problem is that the visible 'disabled' state only updates  
when the event loop runs again. During your long-running method, the  
Cocoa event loop never gets a chance to run. Therefore, any clicks  
are queued and get processed later -- by which point the buttons have  
been enabled again.

You could try disabling the buttons, explicitly redrawing them by  
calling 'display'. You could clear any button clicks from the event  
queue with 'discardEventsMatchingMask:beforeEvent:'.

Re: [Rubycocoa-talk] Array of NSButton ???

[Yvon T. <tho...@fr...>]

Le 22 f=E9vr. 06 =E0 11:37, Jonathan Paisley a =E9crit :

>
> I'm afraid I'm not sure what's going on here. More complete code =20
> would be helpful to understand how things are put together.
>


ok, in Interface Builder all the NSButtons enabled ckeckboxes are set =20=

to false.

then in Controller.rb i do :

	def awakeFromNib
     loadXmlGauche("safari.xml", "BookmarksBar")
		loadXmlDroite("firefox.xml", "Barre personnelle")
     @xmlGauche=3D"safari.xml"
     @xmlDroite=3D"firefox.xml"
     @labelGauche.setStringValue "Safari"
     @labelDroite.setStringValue "Firefox"
     @buttons =3D [@nonVideButtonGauche, @uniqueButtonGauche, =20
@fusionButtonGauche, @loadButtonGauche, @nonVideButtonDroite, =20
@uniqueButtonDroite, @fusionButtonDroite, @loadButtonDroite]
     enableAllButtons                      ####### here i enable all =20
buttons
	end
=09
	def windowDidBecomeMain(sender)
		p "windowDidBecomeMain(sender)"
	end

   def enableAllButtons
     #@nonVideButtonGauche.setHidden true
     #@nonVideButtonGauche.setNeedsDisplay true
     @buttons.each { |b| b.setEnabled(true)}
   end

   def disableAllButtons
     #@nonVideButtonGauche.setNeedsDisplay true
     @buttons.each { |b| b.setEnabled(false)}
   end


after awakeFromNib all the NSButtons are enabled, that' right.

when someone press one of the buttons, saying =20
"Load" (@loadButtonDroite) i do :


	def loadDroite(sender)
     disableAllButtons                     ####### here i disable all =20=

buttons
     loadXmlDroite("firefox.xml", "Barre personnelle")
     @xmlDroite=3D"firefox.xml"
     @labelDroite.setStringValue "Firefox"
     enableAllButtons                      ####### here i re-enable =20
all buttons
   end

i do the "same" (isableAllButtons [...] enableAllButtons) for all the =20=

actions corresponding to all the buttons.


the appearance of the buttons doesn't change and also if, when an =20
action isn't terminated (that's to say between disableAllButtons and =20
enableAllButtons) the click is memorized and, when the action is =20
finished the process enters a new action corresponding to the new =20
button pressed.

if, in between disableAllButtons and enableAllButtons, i press two =20
times a button and another one afterwards, all the click on all the =20
different buttons and the corresponding actions are undertaken each =20
after each...

i'd like to avoid that : no memorization when an action is undertaken.

best,

Yvon=