[pypy-svn] r54183 - pypy/branch/oo-jit/pypy/translator
arigo at codespeak.net
arigo at codespeak.net
Mon Apr 28 19:30:40 CEST 2008
Author: arigo
Date: Mon Apr 28 19:30:39 2008
New Revision: 54183
Added:
pypy/branch/oo-jit/pypy/translator/driver.py.merge.tmp
- copied, changed from r54176, pypy/branch/oo-jit/pypy/translator/driver.py
Log:
merging of svn+ssh://codespeak.net/svn/pypy/branch/jit-hotpath/pypy/translator/driver.py
revisions 51111 to 54176:
------------------------------------------------------------------------
r53562 | fijal | 2008-04-08 04:43:19 +0200 (Tue, 08 Apr 2008) | 2 lines
add --jit-backend to force backend choice
------------------------------------------------------------------------
r53375 | arigo | 2008-04-05 09:36:46 +0200 (Sat, 05 Apr 2008) | 2 lines
Need to mention the ia32 backend here.
------------------------------------------------------------------------
r53096 | arigo | 2008-03-29 17:02:08 +0100 (Sat, 29 Mar 2008) | 8 lines
Biggish refactoring of the way the JitDriver class is used. We're now
supposed to create a (singleton) instance of it, and subclassing
JitDriver is optional. This is somewhat a clean-up. It fixes the
translation issues about the way JIT parameters are set (classmethods
are not really supported in RPython).
All tests (I think) ported to the new style -- keyboard macros rule :-)
------------------------------------------------------------------------
r52917 | arigo | 2008-03-25 16:20:59 +0100 (Tue, 25 Mar 2008) | 2 lines
Be slightly less verbose by default in --jit translated interpreters.
------------------------------------------------------------------------
r52901 | arigo | 2008-03-25 13:15:12 +0100 (Tue, 25 Mar 2008) | 3 lines
fwiw the tests still pass when applying
graphopt.simplify_virtualizable_accesses().
------------------------------------------------------------------------
r52880 | arigo | 2008-03-23 20:08:24 +0100 (Sun, 23 Mar 2008) | 3 lines
Some minimal wacking until the threshold is configurable
at run-time in a pypy-c-jit.
------------------------------------------------------------------------
r52789 | arigo | 2008-03-20 19:35:14 +0100 (Thu, 20 Mar 2008) | 2 lines
Reduce the massive spamming of --jit translations.
------------------------------------------------------------------------
r52727 | cfbolz | 2008-03-19 16:50:52 +0100 (Wed, 19 Mar 2008) | 2 lines
(cfbolz, arigo): use the low level type system bytecode writer
------------------------------------------------------------------------
r52721 | cfbolz | 2008-03-19 14:53:44 +0100 (Wed, 19 Mar 2008) | 2 lines
(arigo, cfbolz): support the hotpath jit stuff in the driver too
------------------------------------------------------------------------
r52360 | arigo | 2008-03-10 17:26:12 +0100 (Mon, 10 Mar 2008) | 8 lines
A branch off the jit-refactoring branch to start playing with
the hotpath approach described in jit-refactoring-plan.txt.
This is all experimental. Some of the changes in this branch might be
easy to merge back if the approach seems to work, but some not. Anyway
for now I will just go ahead and whack at code just to try things out
without worrying about merging.
------------------------------------------------------------------------
r52240 | arigo | 2008-03-07 10:58:28 +0100 (Fri, 07 Mar 2008) | 5 lines
Can't use constant_fold_graph() here because it doesn't preserve
annotations. The original problem was that hannotator.simplify() was
not done any more when using the translator driver.
------------------------------------------------------------------------
r52080 | cfbolz | 2008-03-03 00:32:19 +0100 (Mon, 03 Mar 2008) | 3 lines
fix the driver and translate.py to use the rainbow interpreter instead of the
timeshifter
------------------------------------------------------------------------
r51125 | cfbolz | 2008-01-30 14:07:44 +0100 (Wed, 30 Jan 2008) | 2 lines
a branch for the rainbow interpreter refactoring
------------------------------------------------------------------------
More information about the Pypy-commit
mailing list