[update 1.1] #11194

> Stéphane Ducasse wrote:
>> -  Issue 1932: UTF8TextConverter ??
>> Latin1TextConverter initializeLatin1Encodings.
>> MacRomanTextConverter initializeLatin1Encodings. thanks nicolas/henrik
>
> I still see the same/similar problem.
>
> Try this code in a PharoCore-1.1-11184-UNSTABLE image:
>
> | latin1 utf8 |
> index := 127.
> latin1 := String with: (Character value: index).
> utf8 := latin1 convertToWithConverter: UTF8TextConverter new.
>
> It works. Then try it with the lastest updates loaded. You should get an
> emergency evaluator.
>
> I changed:
>
> ByteString class>>initialize
> "ByteString initialize"
> | latin1 utf8 |
> NonAsciiMap := ByteArray new: 256.
>
> 1 to: 256 do:[:i|
> i <= 128 ifTrue:[
> NonAsciiMap at: i put: 0. "valid ascii subset"
> ] ifFalse:[
> NonAsciiMap at: i put: 1. "extended charset"
>
> ].
> ].
>
> Notice the "i <= 128" vs. "i < 128". That fixes it, but now the concept
> of NonAsciiMap is wrong. I'm not sure what the final fix should be.
>
> --
> Yanni
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> Stéphane Ducasse wrote:
>    
>> -  Issue 1932: UTF8TextConverter ??
>> Latin1TextConverter initializeLatin1Encodings.
>> MacRomanTextConverter initializeLatin1Encodings. thanks nicolas/henrik
>>      
> I still see the same/similar problem.
>
> Try this code in a PharoCore-1.1-11184-UNSTABLE image:
>
> | latin1 utf8 |
> index := 127.
> latin1 := String with: (Character value: index).
> utf8 := latin1 convertToWithConverter: UTF8TextConverter new.
>
> It works. Then try it with the lastest updates loaded. You should get an
> emergency evaluator.
>
> I changed:
>
> ByteString class>>initialize
> "ByteString initialize"
> | latin1 utf8 |
> NonAsciiMap := ByteArray new: 256.
>
> 1 to: 256 do:[:i|
> i<= 128 ifTrue:[
> NonAsciiMap at: i put: 0. "valid ascii subset"
> ] ifFalse:[
> NonAsciiMap at: i put: 1. "extended charset"
>
> ].
> ].
>
> Notice the "i<= 128" vs. "i<  128". That fixes it, but now the concept
> of NonAsciiMap is wrong. I'm not sure what the final fix should be.
>
>    

> On 05.02.2010 18:05, Yanni Chiu wrote:
>    
>> Stéphane Ducasse wrote:
>>
>>      
>>> -  Issue 1932: UTF8TextConverter ??
>>> Latin1TextConverter initializeLatin1Encodings.
>>> MacRomanTextConverter initializeLatin1Encodings. thanks nicolas/henrik
>>>
>>>        
>> I still see the same/similar problem.
>>
>> Try this code in a PharoCore-1.1-11184-UNSTABLE image:
>>
>> | latin1 utf8 |
>> index := 127.
>> latin1 := String with: (Character value: index).
>> utf8 := latin1 convertToWithConverter: UTF8TextConverter new.
>>
>> It works. Then try it with the lastest updates loaded. You should get an
>> emergency evaluator.
>>
>> I changed:
>>
>> ByteString class>>initialize
>> "ByteString initialize"
>> | latin1 utf8 |
>> NonAsciiMap := ByteArray new: 256.
>>
>> 1 to: 256 do:[:i|
>> i<= 128 ifTrue:[
>> NonAsciiMap at: i put: 0. "valid ascii subset"
>> ] ifFalse:[
>> NonAsciiMap at: i put: 1. "extended charset"
>>
>> ].
>> ].
>>
>> Notice the "i<= 128" vs. "i<   128". That fixes it, but now the concept
>> of NonAsciiMap is wrong. I'm not sure what the final fix should be.
>>
>>
>>      
> No, that's me making an off-by-one error :S
> The i<= 128 is correct.
>
>
>    

>>>>
>>
> For that matter, TextConverter initializeLatin1MapAndEncodings should
> probably be reverted as well.
> My mind glitched thinking they were the same (which they are for the
> case I was looking at, with UTF8), but they're only related.
> (I.e. some encodings may or may not have the same mapping as latin1, as
> the old initialize method took care to check...)
>
> Cheers,
> Henry
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

>> i<= 128 ifTrue:[
>> NonAsciiMap at: i put: 0. "valid ascii subset"
>> ] ifFalse:[
>> NonAsciiMap at: i put: 1. "extended charset"
>>
>> ].
>> ].
>>
>> Notice the "i<= 128" vs. "i<  128". That fixes it, but now the concept
>> of NonAsciiMap is wrong. I'm not sure what the final fix should be.
>>
>>    
> No, that's me making an off-by-one error :S
> The i <= 128 is correct.

> Henrik Sperre Johansen wrote:
>    
>>> i<= 128 ifTrue:[
>>> NonAsciiMap at: i put: 0. "valid ascii subset"
>>> ] ifFalse:[
>>> NonAsciiMap at: i put: 1. "extended charset"
>>>
>>> ].
>>> ].
>>>
>>> Notice the "i<= 128" vs. "i<   128". That fixes it, but now the concept
>>> of NonAsciiMap is wrong. I'm not sure what the final fix should be.
>>>
>>>
>>>        
>> No, that's me making an off-by-one error :S
>> The i<= 128 is correct.
>>      
> Then the comments are misleading, because 128 is in the extended charset.
>
>    

> Stéphane Ducasse wrote:
>    
>> Low stress :)
>>
>> Just publish a slice and it will be fixed.
>> I like alpha/unstable/cutting edge just because we are allowed to break the system :)
>>      
> Yep, breaking the system is okay for now. I just set up my build to
> track the updates so I could get a heads-up on possible 1.1 issues,
> while developing on 1.0.
>    

> On 05.02.2010 20:11, Yanni Chiu wrote:
>> Stéphane Ducasse wrote:
>>
>>> Low stress :)
>>>
>>> Just publish a slice and it will be fixed.
>>> I like alpha/unstable/cutting edge just because we are allowed to break the system :)
>>>
>> Yep, breaking the system is okay for now. I just set up my build to
>> track the updates so I could get a heads-up on possible 1.1 issues,
>> while developing on 1.0.
>>
> And a good thing that is :D
>
> In SLICE-OffByOneAndConverterRevert
> (Remember to apply the postscript)
>
> Cheers,
> Henry
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project