Re: [abc-dev] Fwd:

From: Pavel Avgustinov <pavel_at_comlab.ox.ac.uk>
Date: Thu, 11 Sep 2008 15:18:52 +0100

Hi all,

On Thursday 11 September 2008 14:52:01 Eric Bodden wrote:
> Hi all.
>
> 2008/9/11 Chris Allan <mrchrisallan_at_googlemail.com>:
> > I had a very quick look at what was going on, and it seems that when
> > processing the .class files, the field
> > abc.ja.jrag.FieldDeclaration.sootField never gets set, because the
> > conditional at FieldDeclaration line 329 [
> > if(!hostType().getSootClassDecl().declaresFieldByName(name)) ] returns
> > false when looking for the field 'i' - when compiling from source, this
> > expression evaluates to true. I don't know what the significance of this
> > is though, so I gave up investigating! Please let me know if I can
> > provide any more information.

Thanks for this, Chris. It was very helpful in helping us track down the bug!
I and Torbjorn have fixed it in current svn head. Please let us know if you
run into any more trouble!

> To my understanding the error occurs somewhat earlier in the
> compilation process. If I am not mistaken then JastAdd should not even
> touch these files at all because they are read in from bytecode and
> JastAdd is only meant to handle .java files.

Unfortunately, that is a little inaccurate, since ITD weaving happens in the
JastAdd part of the compiler. As such, JastAdd potentially needs to modify
any weavable class file, and so all weavable bytecode is loaded and
jimplified by JastAdd. I'm sorry I didn't correct this misconception when you
mentioned it in response to one of Chris's earlier emails -- it probably
would have saved you some confusion!

Best wishes,
- Pavel
Received on Thu Sep 11 2008 - 15:18:58 BST

This archive was generated by hypermail 2.2.0 : Thu Sep 11 2008 - 16:30:11 BST