[Newcompiler] The decompiler for the old block is ready to test
Marcus Denker
denker at iam.unibe.ch
Sun Nov 11 14:18:53 UTC 2007
On 11.11.2007, at 14:38, Marcus Denker wrote:
>
> On 10.11.2007, at 18:37, Mathieu Suen wrote:
>>
>> Yes , I think but do you know why the decompiler is needed even
>> when you have the ChangeSet file.
>>
>
> The whole stuff in the image is extremely messy...
>
> for one, printing blocks where the printout shows the content of the
> block uses the decompiler.
>
> BlockContext>>decompileString
> ^self decompile decompileString.
>
> This should be rewritten to use the bytecode-offset-to-text mapping
> that is provided by the AST.
I checked, and this is already de-activiated by the NewCompiler
package (the decompileBlock:
in Parser2 just returns nil).
So I made a version of the NewCompiler packge that removes the
FakeDecompiler and put
Decompiler2 as the decompiler to use when a) the "useNewCompiler"
Preference is on
or b) the method is compiled with closures.
Name: NewCompiler-md.278
Author: md
Time: 11 November 2007, 3:11:38 pm
UUID: 399f2191-20df-46ba-88fe-1c5a12ace529
Ancestors: NewCompiler-ms.277
- remove FakeDecompiler
- use Decompiler2 instead
==================================
TODO:
-> fix Block printing to use text + text-to-bytecode mapping
-> either remove #decompileBlock: from old+new decompiler
or implement it for new decompiler.
Marcus
More information about the Newcompiler
mailing list