[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug-kawa] [bug #16388] performance regression
From: |
Per Bothner |
Subject: |
[Bug-kawa] [bug #16388] performance regression |
Date: |
Wed, 19 Apr 2006 22:31:51 +0000 |
User-agent: |
Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.8.0.2) Gecko/20060308 Firefox/1.5.0.2 |
Update of bug #16388 (project kawa):
Assigned to: None => bothner
_______________________________________________________
Follow-up Comment #1:
I suspect this isn't a performance regression in the way you think. Rather
it is that before this code was compiled and now it is interpreted. It's a
new feature: 'let' and 'lambda' forms can now be interpreted, which permits
some macro functionality that wasn't permitted before.
You could verify this by setting the alwaysCompile field in ModuleExp to
true.
That does lead to the *policy* question of when interactively loaded code
should be compiled and when it should be interpreted. Perhaps the default
should be to compile, except perhaps for the very simplest expressions.
"perhaps we should begin adding some performance checks to the test suit"
I have started doing that. See check-inlining in the second-level testsuite
directory.
_______________________________________________________
Reply to this item at:
<http://savannah.gnu.org/bugs/?func=detailitem&item_id=16388>
_______________________________________________
Message sent via/by Savannah
http://savannah.gnu.org/