At 20:11 24.05.00 +0200, Stefan Matthias Aust wrote:
>[Something about benchmarking Interpreter>translate:]

After send out the email, I immediately noticed that of course next is used 
to read in the Smalltalk source for compilation (clearly an issue of 
writing before thinking). I thought that a buffered StandardFileStream 
might have a much better performance and tried to write one.

The result is attached but as this was real brain-surgery I'm not sure 
whether you can file this in - it would modify and probably break the code 
when doing this.  You can at least look at the code and believe my 
profilings :-)   A way to go might be to find the "bufferSize" method and 
change it to 1.  Then everything should be as before.  After a successful 
filein, change it back to 8000 or so.

With buffered read, I get 19% faster translation.  StandardFileStream>>next 
needs only 2.9% or 2.2 seconds now.  I could probably save anther 10% by 
buffering the writing.

bye
-------------- next part --------------
A non-text attachment was scrubbed...
Name: New Changes.1.cs.gz
Type: application/octet-stream
Size: 1784 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20000524/e19e4e0d/NewChanges.1.cs.obj
-------------- next part --------------
--
Stefan Matthias Aust  //  Bevor wir fallen, fallen wir lieber auf