On Nov 4, 2007, at 9:55 AM, Török Edvin wrote:
> On 11/4/07, bugzilla-daemon at cs.uiuc.edu <bugzilla-
> daemon at cs.uiuc.edu> wrote:
>> --- Comment #1 from Chris Lattner <sabre at nondot.org>
2007-11-04
>> 11:45:00 ---
>> The release notes say "The -cee pass is known to be buggy, and may
>> be removed
>> in a future release.". I think we should just do it.
>
> Oops, I should have read the "known problems" section of the
release
> notes, not just the improvements section.
>
> The -cee pass presents interest to me, because I intend to develop a
> static analysis tool, that could use the relationships from -cee.
> [however it would need to support expressions involving more than
> one variable]
> Should I start from scratch, or reuse the "analysis" part of
-cee?
> Could part of -cee be moved to Analysis/ by dropping the code
> transform parts? (if those are the buggy parts).
I actually have a replacement for CEE on my short term todo list. In
addition to not being hopelessly broken, it splits the dataflow fact
analysis out from the transformation pass, so it will probably be more
useful to you than CEE is.
My short-term spare-time priority list is:
1) finish the tutorial
2) finish legalizedagtypes.
3) finish the cee replacement.
-Chris