[Glass] Again, Corrup Error preventing debugging real seaside exception

[Dale Henrichs via Glass]

On 2/1/17 12:57 PM, Mariano Martinez Peck wrote:
>
>
> On Wed, Feb 1, 2017 at 5:50 PM, Dale Henrichs 
> <dale.henrichs at gemtalksystems.com 
> <mailto:dale.henrichs at gemtalksystems.com>> wrote:
>
>     Mariano,
>
>     This second error is not related to the earlier error we were
>     tracking ... I don't see an actual error message in your email, so
>     I'm not exactly sure what is breaking here ...
>
>
> The error message is the one of the screenshot rendered in the browser 
> (attachment named "Screen Shot 2017-02-01 at 5.13.31 PM.png").
That error message was "CorruptObj" and the error message I was looking 
for was the error message from the #changeClassTo: ... The stack that 
you supplied us with was different from the #changeClassTo: stack, as I 
didn't see selfValue on the stack ... doesn't matter though
>
>     Nonetheless, #changeClassTo: was a nasty little performance hack
>     that is no longer needed.
>
>     In GRUtf8CodecStream>>nextPutAll: the following should be done:
>
>       binary
>         ifTrue: [ stream nextPutAll: aString asString ]
>         ifFalse: [ stream nextPutAll: aString _encodeAsUTF8intoString ]
>
>     and in Utf8>>asString the following should be done:
>
>         ^ self decodeToUnicode
>
>     It is odd that this "trick" is failing now (especially without an
>     error message), but I think the right answer is to eliminate the
>     use of #changeClassTo:
>
>
> Thanks!! Those changes DID fix it!!!  Wow...it was hard one this 
> one... I am glad we find it. Do you need anything more from me to open 
> issue or whatever? I will live with overrides for the moment.
Please open an issue against Grease ... up on SeasideST so that we can 
get the fix in ... the #changeTo: hack looks like it went in around 
3.2.2, so I'm not exactly sure when _encodeAsUTF8intoString was added to 
the system (it was added to replace the #changeTo: hack)...
>
> BTW, are you sure about performance? Because I remember that 
> changeClass: hack did improve performance significantly. I wouldn't 
> want to loos that performance boost!
Well, the issue that changeClassTo: was trying to solve was the 
conversion of the Utf8 instance (ByteArray) into a String (required by 
Seaside). #'_encodeAsUTF8intoString' encodes utf8 directly into a String 
instance .... so the performance should be no worse than #changeClassTo:

Dale
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gemtalksystems.com/mailman/private/glass/attachments/20170201/eb8a858c/attachment.html>