markshannon
Follow
markshannon
Follow
Pinned
1,080 contributions in the last year
Activity overview
Contribution activity
May 2022
Created 13 commits in 2 repositories
Created a pull request in python/cpython that received 5 comments
Opened 13 other pull requests in 2 repositories
python/cpython
8
merged
3
open
-
GH-90230: Add stats to breakdown the origin of calls to
PyEval_EvalFrame - GH-91095: Specialize calls to normal python classes
- GH-89914: Push two NULLs (instead of just one) before calls. Preparation for more efficient Python-to-Python call sequence.
- GH-89914: Make the oparg of the YIELD_VALUE instruction equal the stack depth.
- GH-89914: Remove f_globals and f_builtins fields from _PyInterpreterFrame.
-
GH-90690: Remove
PRECALLinstruction - Split refcount stats into 'interpreter' and 'non-interpreter'
- Summarize stats: Increase number shown Predecessor/Successor Pairs from 3 to 5.
- Improve object stats
- GH-92239: Respect PEP 523
- gh-90230: Add stats for freelist use and more allocations.
faster-cpython/ideas
2
merged
Reviewed 8 pull requests in 2 repositories
python/cpython
7 pull requests
- gh-92777: Add LOAD_METHOD_LAZY_DICT
- [WIP] backend detects cold blocks and lays them out at end of function
- [3.10] gh-92311: Let frame_setlineno jump over listcomps
- GH-92239: Respect PEP 523
- gh-91049: Introduce set vectorcall field API for PyFunctionObject
- Fix issues gh-86199 gh-86795
- gh-92154: Expose PyCode_GetCode in the C API
psf/pyperf
1 pull request
Created an issue in python/cpython that received 8 comments
Spurious LINE event when tracing asyncio.tasks.sleep
See nedbat/coveragepy#1368 (comment) for report.
Although this occurs in asyncio.tasks.sleep I suspect that this this is more generally related to …
8
comments
Opened 2 other issues in 2 repositories
hpyproject/hpy
1
open
python/cpython
1
closed
Started 10 discussions in 1 repository
faster-cpython/ideas
faster-cpython/ideas
- Shrinking the inline caches
- Tagged deferred references on the stack
- Virtual iterators
- Making the C stack invisible to bytecode by inserting shim frames
- Reduce reference counting overhead with static analysis and deferred reference counts.
- High level JIT server API
- Broad goals for 3.12 (and beyond)
- Trace-based optimizer
- Faster calling conventions, "hasty call".
- Staying in the interpreter and avoiding calls to C code that call back into Python