Commits

Show all
Author Commit Message Labels Comments Date
Maciej Fijalkowski
optimize it slightly. not look up the dictionary each time we see _digest_size
Branches
rgc-mem-pressure
Justin Peel
merge in default
Branches
rgc-mem-pressure
Justin Peel
improve a test for rgc.add_memory_pressure. not sure that it is really doing what we want it to do.
Branches
rgc-mem-pressure
Justin Peel
small fixes. memory leak is now completely stopped.
Branches
rgc-mem-pressure
Justin Peel
fix hashlib's copy() leak
Branches
rgc-mem-pressure
Maciej Fijalkowski
merge default
Maciej Fijalkowski
merge faster-json branch. It provides encode() optimization for json module, that runs faster than the C extension on CPython (for our benchmark)
Maciej Fijalkowski
make alex happier
Branches
faster-json
Maciej Fijalkowski
don't use identity_dict and remove the commented out stuff
Branches
faster-json
Armin Rigo
Fix: the case marked as "#PyPy" was never actually followed any more. It is doing something better than the other case, so use it again. It also fixes issue917 on the interactive prompt.
Maciej Fijalkowski
a little speedup
Branches
faster-json
Maciej Fijalkowski
implement encode enough so it passes tests
Branches
faster-json
Armin Rigo
Maybe fixes solaris for now.
Armin Rigo
merge heads
Armin Rigo
Change the detection of 'AF_PACKET': instead of detecting it automatically, force it to be True on Linux only. According to http://bugs.python.org/issue8852: The AF_PACKET support was original meant for Linux. When Solaris now also supports that socket family, and with a similar interface, it might be interesting to port that support to Solaris. Notably, "similar" means "but not identical".
Alex Gaynor
convert zip to app-level, it's a tad faster this way. also, more app-level code makes me happy
Alex Gaynor
merged upstream
Alex Gaynor
copy a tad less in rbuilder, only copy the amount of data we have, not the amount we allocated
Armin Rigo
Also add the tests provided on issue917.
Armin Rigo
Fix also the case of '__raw_input__'. Harder to test :-/
Armin Rigo
Test and fix for issue917.
Justin Peel
merge in default
Branches
rgc-mem-pressure
Justin Peel
make CloseHandle (win32) not thread safe to satisfy __del__ conditions
Branches
rgc-mem-pressure
Justin Peel
improve _digest method in hashlib
Branches
rgc-mem-pressure
Maciej Fijalkowski
a test and an obscure fix
Branches
faster-json
Maciej Fijalkowski
use identity_dict and factor out common infrastructure
Branches
faster-json
Armin Rigo
Issue916: test and fix by dhx.
Maciej Fijalkowski
Step one - remove irrelevant clutter, mostly CPython optimizations.
Branches
faster-json
Justin Peel
make sem_close non-threadsafe for now to get __del__ to be acceptable
Branches
rgc-mem-pressure
Armin Rigo
Give the walk_roots operation during a nursery collection its own starting/ending subtime. When running "translate.py --annotate", it is found to take 15% of the time of nursery collections (not really enough to care). For reference, the times are: 12.4% gc-minor including 1.9% gc-minor-walkroots 6.5% gc-collect (i.e. major collections)
  1. Prev
  2. Next