Commits

Show all
Author Commit Message Labels Comments Date
Carl Friedrich Bolz
merge default
Branches
int-tag-untag-as-operations
Carl Friedrich Bolz
make the JIT aware of the downcasts that are present in rtyped-flowgraphs. This allows the JIT to sometimes find out the class of a variable without having to produce a guard_class.
Lukas Diekmann
better name for application representation of list strategies
Lukas Diekmann
added interp_magic method to get the strategy of a list on the application level
Alex Gaynor
simplify this code a bit.
Armin Rigo
Add support for track_allocation=False (the default is True). We can't add many more flags while keeping the same style...
Maciej Fijalkowski
merge
Maciej Fijalkowski
support some combination of flags, I hope it's good enough
Maciej Fijalkowski
raise an exception instead of producing bogus code. Probably would turn some stuff into calls instead of jitcodes
Lukas Diekmann
added some tests for listobject with floats
Carl Friedrich Bolz
merge default
Branches
int-tag-untag-as-operations
Carl Friedrich Bolz
don't use specialize.ll for these calls, but call_location instead. otherwise you cannot use is_constant on high-level objects.
Lukas Diekmann
merge
Lukas Diekmann
added floatstrategy for lists
Armin Rigo
Get rid of 'isstartblock'.
Armin Rigo
Kill dependency on "isstartblock".
Armin Rigo
Chanced upon a "x not in biglist" done in a loop: replace it with "x not in set".
Maciej Fijalkowski
oops, fix translation
Maciej Fijalkowski
copy for slices
Alex Gaynor
fix translation
Armin Rigo
Actually, it seems that the new code replaces any need for the old code. Good, I can remove the comment in rffi.py too...
Armin Rigo
Add a warning message when using stacklet_destroy() with reference counting.
Armin Rigo
Fix.
Armin Rigo
Oups.
Armin Rigo
Fix test.
Armin Rigo
merge heads
Armin Rigo
Fix the test.
Armin Rigo
Add another test that switch() calls are supposed to have an effectinfo of EF_RANDOM_EFFECTS. I thought it was tested already, but it's not, and it fails.
Armin Rigo
Allow at least the front-end to be tested with the "none" gc.
Armin Rigo
Turn off reporting const-fold errors as WARNINGs. This gives quite a bunch of warnings in any C translation, and I've never had any use of them over the years, so well.
  1. Prev
  2. Next