[Rubycocoa-talk] to_ns vs to_nsstring

[Axel M. R. <rub...@ro...>]

Excerpt:
It seems the implementation of string.to_ns is not correct for some 
strings or situations, and to_nsstring works better.

Explanation:
I have come upon a problem which has been bugging me for days. I've 
written a small code example, and that works perfectly, but my main 
code keeps crashing. Let me explain.

I have some code which has worked for months, but after the result 
text got bigger, suddenly stopped working. I've put a lot of debug 
output in it:

# replaceText is a new, highly recursive function
ftext = replaceText(self.mainText,loopStack)
puts "Finished substitutions on Template"
p ftext.class
p ftext.length
joepie= ftext.to_ns
p joepie.class
p joepie.length
self.mainText = joepie

This program doesn't convert the ruby string to NSString, but instead 
returns nil. The output of this is:

Finished substitutions on Template
String
43334
NilClass
2008-11-23 23:40:01.301 Telan[8971:10b] 
TLAppController#genereerPlanning: NoMethodError: undefined method 
`length' for nil:NilClass

As I explained, my small code example works flawlessly with strings 
up till 1000000 chars, so that wasn't the problem. Maybe it's a 
memory problem?

After some more head scratching, I noticed another to_ns method: 
to_nsstring, now deprecated. However, this method works fine!

ftext = replaceText(self.mainText,loopStack)
puts "Finished substitutions on Template"
p ftext.class
p ftext.length
joepie= ftext.to_nsstring
p joepie.class
p joepie.length
self.mainText = joepie

The output of this is (correctly):
Finished substitutions on Template
String
43334
/Users/axel/Develop/svn/telan/TelanRuby/build/Debug/Telan.app/Contents/Resources/TLRTFParse.rb:120:in 
`replaceTemplate': to_nsstring is now deprecated and its use is 
discouraged, please use to_ns instead.
OSX::NSCFString
43333
"TLRTFParse entering observeValueForKeyPath mainText"


So, it seems the method implementation of string.to_ns is different 
from string.to_nsstring, but the method of to_nsstring is less buggy 
than the newer to_ns.

I am not a ruby wizard, and have no idea how to solve this problem 
correctly. As I assume a memory issue, is there a way to 'flush' the 
ruby garbage collector, as you would do with NSAutoreleasePools?
Or is there a way to trace into the ruby interpreter to see where the 
to_ns method fails?


Anyone here on the list who can help me investigate and debug this thing?

thanks


Axel
-- 
_________________________
Axel Roest
axelloroestello@{AIM/MSN} - Skype:axellofono - XOIP: 084-8749988