rubycocoa-talk — talking about RubyCocoa

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 14 mars 06 =E0 14:32, Jonathan Paisley a =E9crit :

>
> Oh well. I still can't understand why the difference. In any case, =20
> avoiding the use of 'include REXML' should fix it.
>

yes i'll know that in a few days, may be hours...

Yvon=

[Rubycocoa-talk] small prob with a menu item

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

Hey all,

i've a menu item called "menuTransformer"

it's one of my ib_outels :

   =20
ib_outlets :mainWindow, :menuChoisir, :menuTransformer, :boutonChoisir, =20=

:boutonTransformer, :messageTextField

in awakeFromNib i do :


   def awakeFromNib
     setMessage("Choisissez un r=E9pertoire...")
     @menuTransformer.setEnabled(false)
   end

here, no prob to set enabled false @menuYtansformer, the app launch =20
normally.

however, this has no effect (the user might continue to choose this =20
menu item)


second prob, when the user has effectively choosen a directory i'd =20
like to set that menu item enabled to true like here :

   def dirChoosen
     setMessage("R=E9pertoire choisi : =AB #{@dir} =BB")
     @transformerBouton.setEnabled(true)
     @transformerBouton.oc_display
     @menuTransformer.setEnabled(true)   ############################ =20=

line 47
     @menuTransformer.oc_display
   end

no prob, whith @transformerBouton (a NSButton) but the application =20
crashes over the line :

@menuTransformer.setEnabled(true)

with the following error message :

/Users/yvon/work/RubyCocoa/AddMenu/build/Development/AddMenu.app/=20
Contents/Resources/Controller.rb:47:in `NSApplicationMain': =20
NSApplicationMain - RBException_NoMethodError - undefined method =20
`setEnabled' for nil:NilClass (OSX::OCException)
	from /Users/yvon/work/RubyCocoa/AddMenu/build/Development/=20
AddMenu.app/Contents/Resources/rb_main.rb:35

althought, @menuTransformer has been seen before in awakeFromNib ???

someone has an idae about that behaviour ???


best,

Yvon

Re: [Rubycocoa-talk] problem with date after standalonification.

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

On 13 Mar 2006, at 7:11, Yvon Thoraval wrote:

> my user did the same experiment over 3 different machines all of  
> them being using Tiger and HFS+

Oh well. I still can't understand why the difference. In any case,  
avoiding the use of 'include REXML' should fix it.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 12 mars 06 =E0 22:15, Jonathan Paisley a =E9crit :

>
> The trace indicates that the first file of your source files being =20
> loaded (after rb_main.rb) is 'FilesToolKit.rb'. It's possible that =20
> some file system differences (perhaps the user has a UFS or case-=20
> sensitive HFS+ filesystem?) cause the order from 'Dir.entries=20
> (path)' in rb_main_init from rb_main.rb to be altered?


my user did the same experiment over 3 different machines all of them =20=

being using Tiger and HFS+

best,

Yvon=

[Rubycocoa-talk] proposal to re-license of RubyCocoa

[FUJIMOTO H. <hi...@fo...>]

hi folks,

I (Fujimoto Hisa, the original author of RubyCocoa) have a plan to
re-license of RubyCocoa about future release.

current licence     : LGPL
future licence      : Ruby's (dual-licnese with LGPL)
re-licensing timing : appropriate timing after building consensus on
                      this (2006.4.1)

If you (as an RubyCocoa contributor) cannot accept this re-licensing
by some reason, please tell me your intention by email to this mailing
list or me.  When there will be no disagreement about this until
0:00GMT of April 1 2006,
the re-licensing will be accepted.  I hope all of you will accept it.

[the background about the re-licensing]

I got an email from an Apple's person a few weeks ago:

  "We (Apple) are interested in including RubyCocoa in a future
  release of Mac OS X. However, the LGPL presents problems for some
  developers who would like to use RubyCocoa to develop their own
  applications.  Would it be possible to relicense RubyCocoa under a
  different license?"

It's the purpose of the re-licensing. It will make distribution of a
software which depends on RubyCocoa easily for some developers.
hope again all of you will accept this.

thanks,
--
hisa

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 12 mars 06 =E0 22:24, Jonathan Paisley a =E9crit :

>
> Some extra info on this - UFS seems to return files in creation =20
> order, whereas HFS+ returns files in alphabetical order.
>

at that time it was quit normal to get 'FilesToolKit.rb at first =20
position.

because i had to write a catalog.xml file (then use REXML) according =20
to the folders where the dtds are lying, before any use of an xml =20
document asking for an outside dtd. i had a bad experience with =20
plist's dtd when the Apple site was unreachable, the application had =20
to wait...(roughly 9 in the morning, Paris hour)

right now i do that in Controller#initialize (your advice using yaml =20
(then Date class) before REXML).

for the file system, i've asked to him (it's around midnight now in =20
france) BUT, i think he is careful and, afaik, MacOS X on UFS file =20
system as really bad reputation, i'll know that tomorrow morning.

best,

Yvon=

Re: [Rubycocoa-talk] problem with date after standalonification.

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

On 12 Mar 2006, at 21:15, Jonathan Paisley wrote:

> The trace indicates that the first file of your source files being  
> loaded (after rb_main.rb) is 'FilesToolKit.rb'. It's possible that  
> some file system differences (perhaps the user has a UFS or case- 
> sensitive HFS+ filesystem?) cause the order from 'Dir.entries 
> (path)' in rb_main_init from rb_main.rb to be altered?

Some extra info on this - UFS seems to return files in creation  
order, whereas HFS+ returns files in alphabetical order.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

On 12 Mar 2006, at 20:43, Yvon Thoraval wrote:

>
> Le 12 mars 06 =E0 13:32, Jonathan Paisley a =E9crit :
>
>> The output of 'kdump | grep NAMI' after running under 'ktrace' =20
>> would help trace it down.
>
> Is that ok  ?

The trace indicates that the first file of your source files being =20
loaded (after rb_main.rb) is 'FilesToolKit.rb'. It's possible that =20
some file system differences (perhaps the user has a UFS or case-=20
sensitive HFS+ filesystem?) cause the order from 'Dir.entries(path)' =20
in rb_main_init from rb_main.rb to be altered?

Does the ktrace output from running it on your computer indicate the =20
same file being loaded first? (if you don't want to look at the =20
ktrace output you could add 'puts' statements to the top of each of =20
your ruby source files).

If Dir.entries is the culprit you could try changing it to =20
'Dir.entries(path).sort' to see if that makes any difference.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

On 12 Mar 2006, at 19:38, Yvon Thoraval wrote:

>> Again, I'd recommend not using 'include REXML' at all.
>
> then, not to use REXML at all you mean ?

No, using 'include REXML' is a shortcut so you can reference REXML's  
'Document' class and others without explicitly qualifying with  
REXML::. It has the unfortunate side-effect that REXML::Date becomes  
a top-level constant.

You can still use REXML, just ensure that you reference things as  
REXML::Document REXML::Element etc rather than just plain Document or  
Element.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 12 mars 06 =E0 13:32, Jonathan Paisley a =E9crit :

>
>> that's new, because i need to build a catalog.xml file at start, =20
>> i'll a llok on how to move it elsewhere.
>
> Again, I'd recommend not using 'include REXML' at all.

then, not to use REXML at all you mean ?

what would you suggest instead of ?

>
>> but this doesn't explain why it works on my computer not on user's =20=

>> one ?
>
> Yes, it is very strange. The output of 'kdump | grep NAMI' after =20
> running under 'ktrace' would help trace it down.
>

ok i'll ask that to the user may be he has enough time and want to =20
learn ;-)

best,

Yvon=

Re: [Rubycocoa-talk] problem with date after standalonification.

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

On 12 Mar 2006, at 12:14, Yvon Thoraval wrote:

> that's new, because i need to build a catalog.xml file at start,  
> i'll a llok on how to move it elsewhere.

Again, I'd recommend not using 'include REXML' at all.

> but this doesn't explain why it works on my computer not on user's  
> one ?

Yes, it is very strange. The output of 'kdump | grep NAMI' after  
running under 'ktrace' would help trace it down.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 12 mars 06 =E0 12:35, Jonathan Paisley a =E9crit :

> That would probably explain it, if you have used 'include REXML' at =20=

> before including 'date'.
>
> For example:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
> require 'rexml/document'
> include REXML
> require 'date'
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D
>
> Gives "Date is not a class (TypeError)".
>
> Try ensuring that 'date' is required before 'rexml/document' in =20
> your code, or -- better -- remove 'include REXML' from the top level.
>


that's the case, indirectly (because i don't use directly only yaml =20
use it) here is my actual order of requirement :

require 'rexml/document'
include REXML
buildCatalog

require 'osx/cocoa'
require 'iconv'
require 'xml/libxml'
require 'xml/xslt'
require 'yaml'

that's new, because i need to build a catalog.xml file at start, i'll =20=

a llok on how to move it elsewhere.

but this doesn't explain why it works on my computer not on user's one ?

anyway, the best is to find a workaround...

best,

Yvon=

Re: [Rubycocoa-talk] problem with date after standalonification.

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

> module REXML
> 	Copyright =3D "Copyright =A9 2001, 2002, 2003, 2004 Sean Russell =
=20
> <se...@ge...>"
> 	Date =3D "2005/224"
> 	Version =3D "3.1.3"
> end

That would probably explain it, if you have used 'include REXML' at =20
before including 'date'.

For example:
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D
require 'rexml/document'
include REXML
require 'date'
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D

Gives "Date is not a class (TypeError)".

Try ensuring that 'date' is required before 'rexml/document' in your =20
code, or -- better -- remove 'include REXML' from the top level.

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 12 mars 06 =E0 01:11, Jonathan Paisley a =E9crit :

>
> The above is all ok and ends up with one class Foo with two methods.

yes, in fact, i knew that...

>
> The exception you're getting (Date is not a class) comes if the =20
> name 'Date' has previously been used for another purpose, for =20
> example as a module:
>
>   module Date
>   end
>   class Date
>   end #=3D> Date is not a class (TypeError)
>
> Can you check that this doesn't happen anywhere in your code or any =20=

> library code that's getting used?

i did a grep over the Resources folder, the only thing suspectable is =20=

in rexml :

 > grep -nIrw Date *
[...]
ThirdParty/rexml/rexml.rb:24:   Date =3D "2005/224"
[...]

the whole file being :

module REXML
	Copyright =3D "Copyright =A9 2001, 2002, 2003, 2004 Sean Russell =
=20
<se...@ge...>"
	Date =3D "2005/224"
	Version =3D "3.1.3"
end

but may be it's not a prob (i'm coming from java side) assigning a =20
string to what is supposed to be a class...

nothing more when greping over the whole Contents of this app.

best,

Yvon=

Re: [Rubycocoa-talk] problem with date after standalonification.

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

>>
>> format.rb line 6 being :
>> class Date
>>
>> date.rb line 197 :
>> require 'date/format'
>
> i think their is a prob here because at line 392 of date.rb you see :
>
> class Date
>
> then two class Date are defined and interrelated ???

That's ok, since in ruby classes are open, so you can redefine them  
later. e.g.

   class Foo
     def method1
     end
   end

   class Foo
     def method2
     end
   end

The above is all ok and ends up with one class Foo with two methods.

The exception you're getting (Date is not a class) comes if the name  
'Date' has previously been used for another purpose, for example as a  
module:

   module Date
   end
   class Date
   end #=> Date is not a class (TypeError)

Can you check that this doesn't happen anywhere in your code or any  
library code that's getting used?

Re: [Rubycocoa-talk] problem with date after standalonification.

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

Le 11 mars 06 =E0 23:45, Yvon Thoraval a =E9crit :

>
> format.rb line 6 being :
> class Date
>
> date.rb line 197 :
> require 'date/format'
>
>


i think their is a prob here because at line 392 of date.rb you see :

class Date


then two class Date are defined and interrelated ???

Yvon=

[Rubycocoa-talk] problem with date after standalonification.

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

Hey all,

i'm using yaml on my code, the yaml code depends upon date this cause  
a crash when starting a RubyCicia app one a computer of one of my  
users, not on my computer here is the error from console :

/Applications/BookmarksMerge.app/Contents/Resources/ThirdParty/date/
format.rb:6: Date is not a class (TypeError)
	from /Applications/BookmarksMerge.app/Contents/Resources/ThirdParty/
date.rb:197
	from /Applications/BookmarksMerge.app/Contents/Resources/ThirdParty/
yaml/rubytypes.rb:2
	from /Applications/BookmarksMerge.app/Contents/Resources/ThirdParty/
yaml.rb:392
	from /Applications/BookmarksMerge.app/Contents/Resources/rb_main.rb:46


format.rb line 6 being :
class Date

date.rb line 197 :
require 'date/format'

yaml/rubytypes.rb:2 :
require 'date'

yaml.rb:392 :
require 'yaml/rubytypes'


i never get that on my computer even if i launch the standaloneify  
version of the app....

strange this message stating date isn't a class pointing the line  
where it is defined as a class...

best,

Yvon

Re: [Rubycocoa-talk] libxml and standalonification

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

Le 9 mars 06 =E0 09:17, Jonathan Paisley a =E9crit :

>
> Where did this library come from? Are you using the latest libxml =20
> ruby gem?

no the install using rubygems didn't work because rubygems wasn't =20
able to find iconv and related.
even if iconv has been re-installed by dp for ruby.

let me tell you the story :

i did a complete re-install of :
ruby, rubycocoa, rubygems

using darwinports


ive also installed by hand  rubyaesoa (afaik no gem for it)

then, because the install of the libxml binder fails using rubygems, =20
i did that "by hand" using :

ruby ext/xml/extconf.rb  --with-iconv-dir=3D/usr
make
sudo make install

(advice given to me on comp.lang.ruby, i realize now the arg "--with-=20
iconv-dir=3D/usr" might be wrong because the right iconv for the ruby =20=

installed in /opt, might be also installed by dp under /opt ?)

then the ruby used to install libxml is the latest installed by dp (i =20=

did change my $PATH to get /opt/local/bin at first position)

and i'm able to use it from terminal and also to put a require of it =20
in a rubycocoa app.

>
>> /Library/Frameworks/libxml.framework
>> /Library/Frameworks/libxslt.framework
>
> I'm confused - libxml and libxslt are installed as standard as =20
> normal libraries in /usr/lib. Where did these frameworks come from?

from GnomeLibXmlCombo-2005-11-06 which contains :

libexslt.framework
libxml.framework
libxslt.framework
xmllint
xsltproc

i've installed that because i didn't know, at that time, than libxml =20
is standard on Tiger.

Re: [Rubycocoa-talk] libxml and standalonification

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

On 9 Mar 2006, at 6:53, Yvon Thoraval wrote:

> require 'xml/libxml'

Where did this library come from? Are you using the latest libxml  
ruby gem?

> /Library/Frameworks/libxml.framework
> /Library/Frameworks/libxslt.framework

I'm confused - libxml and libxslt are installed as standard as normal  
libraries in /usr/lib. Where did these frameworks come from?

[Rubycocoa-talk] libxml and standalonification

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

Hey all,

i've an app making use of :

require 'xml/libxml'

the app crash at startup with the following message :

> dyld: Library not loaded: @executable_path/../Frameworks/
> libxml.framework/Versions/2.6.22/libxml
>   Referenced from: /Applications/BookmarksMerge.app/Contents/MacOS/
> BookmarksMerge
>   Reason: image not found
> Mar  8 20:39:01 G3-Kaboss crashdump[1848]: BookmarksMerge crashed
> Mar  8 20:39:01 G3-Kaboss crashdump[1848]: crash report written to: /
> Users/jo/Library/Logs/CrashReporter/BookmarksMerge.crash.log

the user got the same error message after commenting out the line :

require 'xml/libxml'

(this line was only for testing purpose, i don't use ruby-libxml at  
this time, but plan to)

the binder bundle "libxml.bundle" is at the right place :  .../ 
Contents/lib/xml

i know the prob isn't coming from standaloneify.rb.

i've to put by myself all of the stuff related to libxml2, but don't  
know exactly what and where.

first i would put a copy of :

/Library/Frameworks/libxml.framework
/Library/Frameworks/libxslt.framework

into : .../Contents/Frameworks

second  :

/usr/lib/libxml2.2.dylib
/usr/lib/libxml2.dylib
/usr/lib/libxml2.la

/usr/lib/libxslt.1.dylib
/usr/lib/libxslt.dylib
/usr/lib/libxslt.la

into : .../Contents/lib

i don't think i'd have to copy the headers : /usr/include/libxml2

do you think that' enough, at the right place ?

does i have to change the $LOAD_PATH of ruby, or, given the  
directories are the same as for ruby and libxml bundle, nothins to  
change there ?

best,

Yvon

Re: [Rubycocoa-talk] NSAlert (sheet) and contextInfo

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

Le 6 mars 06 =E0 11:31, Jonathan Paisley a =E9crit :

> I think it's probably easiest to ignore the contextInfo argument. =20
> Since you know which object is going to be the delegate, just set =20
> an instance variable on the delegate.  Assuming the delegate is the =20=

> same as the caller of beginSheetModalForWindow:... then an example =20
> might be:
>
>   @done_alert_proc =3D proc { saveFile("thisFileName") }
>   alert.beginSheetModalForWindow_...(...)
>
> def alertDidEnd:...
>   # Check return code first, then:
>   @done_alert_proc.call
> end
>


ok, much more clever and easier to do ))

Yvon=

Re: [Rubycocoa-talk] OSX::NSAppleEventDescriptor.data

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

Le 6 mars 06 =E0 11:16, Jonathan Paisley a =E9crit :

> Try something like this:
>
> keyDirectObject =3D '----'.unpack('N')[0]
> arg =3D event.descriptorForKeyword(keyDirectObject).stringValue.to_s
>


yes, Luc Heinrich gave me also this advice, that's working now.

Yvon=

Re: [Rubycocoa-talk] NSAlert (sheet) and contextInfo

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

On 5 Mar 2006, at 20:09, Yvon Thoraval wrote:

> i had a look about the class of the returned contextInfo giving :
>
> contextInfo.class = OSX::ObjcPtr  (presumably an Obj-C Pointer) ?

Since the C type declaration is 'void*', RubyCocoa doesn't know what  
type it really is. It wraps the value in the ObjcPtr class.

> my first try used a string as a contextInfo : "this is the  
> contextInfo".
>
> the value returned by contextInfo.to_s was : #<OSX::ObjcPtr:0x3b3f510>

The ObjcPtr is pointing to a C string.

> the better for me would be to pass a method with it's associated  
> args for example :
>
> saveFile("thisFileName")
>
> how could i implement that with contextInfo ???

I think it's probably easiest to ignore the contextInfo argument.  
Since you know which object is going to be the delegate, just set an  
instance variable on the delegate.  Assuming the delegate is the same  
as the caller of beginSheetModalForWindow:... then an example might be:

   @done_alert_proc = proc { saveFile("thisFileName") }
   alert.beginSheetModalForWindow_...(...)

def alertDidEnd:...
   # Check return code first, then:
   @done_alert_proc.call
end

Re: [Rubycocoa-talk] OSX::NSAppleEventDescriptor.data

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

Sounds like you need to read up on how the AppleEvent manager works,  
which is independent of RubyCocoa.

Try something like this:

keyDirectObject = '----'.unpack('N')[0]
arg = event.descriptorForKeyword(keyDirectObject).stringValue.to_s

Re: [Rubycocoa-talk] OSX::NSAppleEventDescriptor.data

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

Le 6 mars 06 =E0 10:23, Yvon Thoraval a =E9crit :

> i think i have something to do with unpack ?
>


unpack couldn't be a solution because when trying all of :

['A', 'a', 'B', 'b', 'H', 'h', 'M', 'm', 'u', 'Z']

as a format i never get back the ten zeros...

Yvon=