In the weblogic session stuff, most of the generated bytecode methods were done 
on stub classes used only for compilation.  I actually put the methods there 
and left them there, with references in Javadoc to what method was used to 
generate the code.  This can't normally be done, but I think the idea of 
keeping the dummy method used to generate the starting code seems pretty 
critical for maintenance.  

+1 for the blog post (AND a wiki page that can be updated beyond the life of 
the blog).


----- Original Message -----
From: "Geert Bevin" <[EMAIL PROTECTED]>
To: [email protected]
Sent: Tuesday, October 2, 2007 1:14:30 PM (GMT-0600) America/Chicago
Subject: Re: [tc-dev] Line numbers in generated code?


On 02 Oct 2007, at 20:00, Tim Eck wrote:

> Couldn't agree more with regards to meaningful label names. Another
> pointer is to leave commented psudeo-code for anything more  
> complicated
> than simple getter/setter.

Yup, either that, or split the bytecode up in blocks that do specific  
actions and comment those. Just leaving whole pages of pasted  
bytecode instructions, without any blank lines or comments is ...  
horrible. The code takes way too long to read during maintenance  
(going through that with the CHM stuff atm :-/).

Actually, I'm gonna create an internal wiki page with ASM do's and  
don'ts ... I think that could be useful ... or should that be  
public ... with a blog post. Hmmm that's maybe better.

--
Geert Bevin
Terracotta - http://www.terracotta.org
Uwyn "Use what you need" - http://uwyn.com
RIFE Java application framework - http://rifers.org
Music and words - http://gbevin.com

_______________________________________________
tc-dev mailing list
[email protected]
http://lists.terracotta.org/mailman/listinfo/tc-dev

_______________________________________________
tc-dev mailing list
[email protected]
http://lists.terracotta.org/mailman/listinfo/tc-dev

Reply via email to