Re: is baf really first?

[Patrick LAM <plam@sable.mcgill.ca>]

On Fri, 1 Mar 2002, Navindra Umanee wrote:

> According to the "famous Soot picture", Soot first produces Baf from
> bytecode and then produces Jimple from Baf.
>
> So why is it when I run Soot with the verbose option, it seems that
> Jimple is being produced first?  A glance at the Coffi code seems to
> confirm this.

Baf is conceptually first, but we don't have a bytecode-to-Baf path, while
we do have a bytecode-to-Coffi and a Coffi-to-Jimple path.  This is what
we use.  So we say that we produce Baf, but we don't really do any such
thing.

There's also a Jimple parser, of course.

pat