aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/python/matplotlib/py2/src/_path_wrapper.cpp
diff options
context:
space:
mode:
authorarkady-e1ppa <arkady-e1ppa@yandex-team.com>2024-05-20 12:25:58 +0300
committerarkady-e1ppa <arkady-e1ppa@yandex-team.com>2024-05-20 12:39:26 +0300
commitd2a413e97eb8e248a266f0129f6f8fecbc8605b7 (patch)
tree38ce318e38bbd14d578a65f4bb9809c3ff5d689c /contrib/python/matplotlib/py2/src/_path_wrapper.cpp
parent73d783fe0b096861111e19b63fa2259296050916 (diff)
downloadydb-d2a413e97eb8e248a266f0129f6f8fecbc8605b7.tar.gz
YT-21402: Fibers refactoring pt.2: Slightly better encapsulation of fiber introspection part and moved fls back to switch handler
In this pr we refactor fiber state machine so that it is more clear what its purpose is: it protects Fls (and some other fields) across fiber-introspector interaction over the fiber life-cycle and also fix some problems related to the shutdown sequence: -Fls is back to being created at FiberSwitchHandler (although FiberSwitchHandler is still inappropriately placed in the same UT as fiber scheduler, but that issue is intended to be fixed in later prs). -FiberId generation is also back to FiberSwitchHandler, with its value being cached inside fiber for introspector use. This shouldn't make things worse, but would make it clearer who *owns* the FiberId and who *views* it. -TFiberRegistry now uses shutdown callback to clear the last batch of fibers instead of doing so in dtor, which is apparently never called. 6f8c7b1c741b775128a3a7a5aeaefbb2563fcf5d
Diffstat (limited to 'contrib/python/matplotlib/py2/src/_path_wrapper.cpp')
0 files changed, 0 insertions, 0 deletions