@HHVMCommits
 Menlo Park, CA
hhvm.com/repo
Joined July 2014
  • Tweets20.2K
  • Following0
  • Followers34
[hhvm] github.com/facebook/hhvm/com… Bin Liu - include function order when serializing profile data Summary: Avoid hfsort to make deserializat...

include function order when serializing profile data · facebook/hhvm@43f7eaf

Summary: Avoid hfsort to make deserialization faster. It also gives a stable order, even when hfsort isn't stable. Reviewed By: markw65 Differential Revision: D13855770 fbshipit-source-...

github.com
[hhvm] github.com/facebook/hhvm/com… Emil Hesslow - Remove NumArgs from array_merge and array_merge_recursive Summary: - Using NumArgs here wa...

Remove NumArgs from array_merge and array_merge_recursive · facebook/hhvm@2db3f19

Summary: - Using NumArgs here was most likely an optimization - IterateV should be faster. So even if it is a regression it should be tiny. Reviewed By: paulbiss Differential Revision: D1377575...

github.com
[hhvm] github.com/facebook/hhvm/com… Ted Spence - Improve findRefs to capture class_meth without apostrophes Summary: Modify the `find_refs` t...

Improve findRefs to capture class_meth without apostrophes · facebook/hhvm@783305c

Summary: Modify the `find_refs` task to be aware of function references that use apostrophes. This occurs in a few types of cases: * When using fun('global') to refer to a global f...

github.com
[hhvm] github.com/facebook/hhvm/com… Philip Jameson - Add load() statements to TARGETS files Reviewed By: styurin Differential Revision: D137...

Add load() statements to TARGETS files · facebook/hhvm@2fe76e5

Reviewed By: styurin Differential Revision: D13733551 fbshipit-source-id: 88050c315c8388cff9b1807bd83f4bfa50ca16b9

github.com
[hhvm] github.com/facebook/hhvm/com… Bill Fumerola - Use Iterate{K,}V in socket_select() helpers Reviewed By: alexeyt Differential Revision: ...

Use Iterate{K,}V in socket_select() helpers · facebook/hhvm@c89fa79

Reviewed By: alexeyt Differential Revision: D13525557 fbshipit-source-id: 497beb007c3420e16aaab08faae00dd8155c281c

github.com
[hhvm] github.com/facebook/hhvm/com… Bill Fumerola - migrate away from unspecialized arrays: ext_sockets Summary: HNI not changed: - `socket_c...

migrate away from unspecialized arrays: ext_sockets · facebook/hhvm@3666a75

Summary: HNI not changed: - `socket_create_pair`, `stream_socket_pair`: assigned by ref without type info - `stream_socket_pair`: returns `mixed` (`false|array`) Reviewed By: ricklavoie Different...

github.com
[hhvm] github.com/facebook/hhvm/com… Bill Fumerola - migrate away from unspecialized arrays: ext_stream Summary: HNI not changed: - `stream_so...

migrate away from unspecialized arrays: ext_stream · facebook/hhvm@b92b95e

Summary: HNI not changed: - `stream_socket_pair`: returns mixed (false|array) HNI changed to `varray`: - `stream_get_wrappers` Reviewed By: ricklavoie Differential Revision: D13490876 fbshipit-...

github.com
[hhvm] github.com/facebook/hhvm/com… Maggie Moss - Move gconst initalization check from typing to tast_check (8/n) Summary: Ask: Removing this...

Move gconst initalization check from typing to tast_check (8/n) · facebook/hhvm@ccebd95

Summary: Ask: Removing this check did not cause any tests to fail, I am wondering if there's ever a scenario where a const would be uninstantiable? Reviewed By: kmeht Differential Revisio...

github.com
[hhvm] github.com/facebook/hhvm/com… Paul Bissonnette - Fix a SetArray insertion bug Summary: We were growing the keyset even when insertion w...

Fix a SetArray insertion bug · facebook/hhvm@0f93a26

Summary: We were growing the keyset even when insertion was going to fail, potentially moving the underlying keyset and leaving us with a pointer into freed memory. Reviewed By: alexeyt Different...

github.com
[hhvm] github.com/facebook/hhvm/com… Paul Bissonnette - Fix the catch trace for RaiseHackArrParamNotice in irgen-builtin Summary: Without this...

Fix the catch trace for RaiseHackArrParamNotice in irgen-builtin · facebook/hhvm@5914fbd

Summary: Without this catch we would could end up throwing without a valid pre-live actrec on the stack. Reviewed By: ricklavoie Differential Revision: D13893298 fbshipit-source-id: 3b0643487791...

github.com
[hhvm] github.com/facebook/hhvm/com… Andrew Kennedy - New inference: remove even more tyvars plumbing Summary: This is the final part of the n...

New inference: remove even more tyvars plumbing · facebook/hhvm@27796c0

Summary: This is the final part of the new approach to fresh tyvar management, removing the extra "tyvar set" parameter and return value from `fresh_unresolved_type_add_tyvars` an...

github.com
[hhvm] github.com/facebook/hhvm/com… Andrew Kennedy - New inference: remove more tyvars plumbing Summary: Remove various dead parameters and p...

New inference: remove more tyvars plumbing · facebook/hhvm@e53c5c1

Summary: Remove various dead parameters and paths after previous redesign of tyvar generation. Reviewed By: manzyuk Differential Revision: D13886131 fbshipit-source-id: 48c7f64e5f927dae426764212...

github.com
[hhvm] github.com/facebook/hhvm/com… Andrew Kennedy - New inference: simplify tyvars plumbing Summary: Major refactoring of our tyvars plumbin...

New inference: simplify tyvars plumbing · facebook/hhvm@ea49af6

Summary: Major refactoring of our tyvars plumbing. Instead of passing around "freshly generated type variables" explicitly, we store sets of type variables on a stack in `env`, pu...

github.com
[hhvm] github.com/facebook/hhvm/com… Sasha Manzyuk - Eagerly solve the type of an unpacked function argument when enforcing it is unpackable S...

Eagerly solve the type of an unpacked function argument when enforcin… · facebook/hhvm@f097255

…g it is unpackable Summary: We do a case analysis on the type of an unpacked argument to enforce it is unpackable; in particular, if the type is a tuple, we check that the types of the components...

github.com
[hhvm] github.com/facebook/hhvm/com… Kasper Kopec - Revert "D13225280: Limit parser thread count to only physical cores" Summary: With fix fro...

Revert "D13225280: Limit parser thread count to only physical cores" · facebook/hhvm@4a670b7

Summary: With fix from D13884788, this optimization no longer impacts parsing memory in significant way, while (according to Mark) having negative performance on time of subsequent hhbbc phase. Re...

github.com
[hhvm] github.com/facebook/hhvm/com… Kasper Kopec - Avoid passing autoload map data from HHVM to HackC process, reduce peak build memory by 50%...

Avoid passing autoload map data from HHVM to HackC process, reduce pe… · facebook/hhvm@1c24860

…ak build memory by 50% Summary: Having impact in HackC codebase is easier than I suspected! All the peak memory was wasted deserializing an enormous json blob in each of the workers, with the big...

github.com
[hhvm] github.com/facebook/hhvm/com… Tatiana Racheva - Fix test_mergedriver_finishes_quickly_hack_already_running Summary: The issue was that ...

Fix test_mergedriver_finishes_quickly_hack_already_running · facebook/hhvm@8ac612e

Summary: The issue was that the test files started with whitespace, but we require (as of when?) that

github.com
[hhvm] github.com/facebook/hhvm/com… Guilherme Ottoni - Get rid of a warning regarding output buffer used by snprintf Reviewed By: paulbiss D...

Get rid of a warning regarding output buffer used by snprintf · facebook/hhvm@d425a4b

Reviewed By: paulbiss Differential Revision: D13830216 fbshipit-source-id: 863211f8235473e7f19e2234cb94040839a6a978

github.com
[hhvm] github.com/facebook/hhvm/com… Bin Liu - delete healthlevel from server stats Summary: The only way this can be observed is by sending a...

delete healthlevel from server stats · facebook/hhvm@bc3a0f0

Summary: The only way this can be observed is by sending an admin-port command /stats.kvp?keys=health_level. The value is almost always 0 ("Bold"). When it isn't, knowing...

github.com
[hhvm] github.com/facebook/hhvm/com… Kunal Mehta - Remove Full_fidelity_syntax_tree.from_root Summary: The way we handle the file mode in the ...

Remove Full_fidelity_syntax_tree.from_root · facebook/hhvm@8edd693

Summary: The way we handle the file mode in the FFP is mildly clowny: the mode information lives in the parser env, but nothing guarantees that the env that you pass into `Full_fidelity_parser.pars...

github.com
Load more