Hello all,

Here is an update on the merge of the open64-booster branch into the
trunk that we started last month.

We have completed all of the testing described in the check-in policy
posted at http://wiki.open64.net/index.php/CheckinPolicy.  No
regressions were found and the results of some of the test suites were
improved.

We also received review comments from some gatekeepers and external
reviewers, which I summarize here:

o In the changes to the inlining heuristics, the reviewer pointed out
some dead code that had been added during development and testing.
The attached patch addresses the issue by removing that code.

o In the review of the code generator changes, there was some
discussion about whether some new properties should be at the
"processor" level instead of the "isa" level.  The developer felt that
these properties were easier to use at the "isa" level given the
current design, where only scheduling information is
processor-specific.  Also, there were some formatting changes where
the developer made the formatting more consistent across the file --
any remaining inconsistencies are unintentional and will be addressed
in a future update.

o In the optimizer changes, the reviewer asked whether we had
considered doing the LOWER_SIMPLIFY_BIT_OP transformation earlier, in
preopt_phase instead of mainopt_phase.  The developer replied that the
main reason to do the transformation while lowering was to avoid
complexity in the implementation.  There were also some specific code
logic questions that were resolved.

Thanks to the reviewers for their feedback!

With the approval of one of the global gatekeepers, we would now like
to commit the merge changes to the trunk.

Finally, I want to add that we understand the difficulty of making
large merges through the gatekeeper review system now in place.  We're
taking steps that we hope will result in more individual commits and
smaller merges in the future.

-David Coakley / AMD Open Source Compiler Engineering

Attachment: inline_cleanup.diff
Description: Binary data

------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
Open64-devel mailing list
Open64-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/open64-devel

Reply via email to