Commits

Author Commit Message Labels Comments Date
Greg Price
signature: support pointer types
Greg Price
cut dependency annotation.model -> unaryop, binaryop This makes annotation.model very lean on dependencies, which keeps rlib.types lean, which will be very helpful in importing the latter all over for signatures without creating cyclic dependencies.
Greg Price
signatures: basestring type
Greg Price
malloc: small cleanups Some logic and an error message hadn't kept up with changes over the years.
Philip Jenvey
redundant
Philip Jenvey
minor cleanup
Branches
py3k
Philip Jenvey
use fixedview here
Branches
py3k
Philip Jenvey
skip frozen module/imp.init_builtin/objspace.lonepycfiles=False tests
Branches
py3k
Philip Jenvey
merge default
Branches
py3k
Philip Jenvey
invoke java in headless mode so it's less annoying on osx
Philip Jenvey
whitespace, arrange imports
Philip Jenvey
apply the TestOOtype skips to jvm/cli explicitly string_to_float seems to work whereas parts_to_float doesn't
Antonio Cuni
force the raw buffer if we detect an invalid read
Branches
virtual-raw-mallocs
Antonio Cuni
force the raw buffer if we detect an invalid write
Branches
virtual-raw-mallocs
Antonio Cuni
one more overlapping case to detect
Branches
virtual-raw-mallocs
Antonio Cuni
improve the test
Branches
virtual-raw-mallocs
Antonio Cuni
implement forcing of the virtual raw memory
Branches
virtual-raw-mallocs
Antonio Cuni
use the new RawBuffer to implement the virtual raw memory
Branches
virtual-raw-mallocs
Antonio Cuni
keeping track of the lengths of writes is not enough: we want to make sure that also the descrs are the same, in case e.g. I store a long and then read a double (in that case, we just abort the optimization)
Branches
virtual-raw-mallocs
Antonio Cuni
implement a "raw virtual buffer": the idea is that we can write values to the buffer at arbitrary positions in the buffer: as long as we write and read to non-overlapping pices of memory, it's all fine, but we need to detect the case in which a write might partially overwrite the memory already stored earlier: in that case, we raise an exception and abort the optimization
Branches
virtual-raw-mallocs
Armin Rigo
hg merge default
Branches
remove-globals-in-jit
Armin Rigo
Fix
Branches
stm-thread-2
Armin Rigo
Fix
Branches
stm-thread-2
Antonio Cuni
first passing test: very simple and lots of things still to do (starting from forcing)
Branches
virtual-raw-mallocs
Antonio Cuni
rewrite the failing test to use an untyped buffer and {get,set}arrayitem_gc ops, because this is what happends in cffi
Branches
virtual-raw-mallocs
Antonio Cuni
a failing test
Branches
virtual-raw-mallocs
Antonio Cuni
a branch where to add support for virtual raw mallocs: the goal is to optimize away the temp buffer created to do cffi calls
Branches
virtual-raw-mallocs
Armin Rigo
Typo in C sources, which only gave warnings.
Armin Rigo
Left behind in previous merge
Branches
stm-thread-2
Armin Rigo
hg merge default
Branches
stm-thread-2
  1. Prev
  2. Next