You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Bc new users with fancy hw keep asking about it and i suppose once
we add support the project will prolly get popular with them "career
hackers" with "nice furniture" 😹
Immediately known (now solved) issues,
our use of XDG_RUNTIME_DIR
on linux in the UDS tpt backend
for the selection of the UDSAddress.def_bindspace is problematic.
see this SO answer which
indicates using DARWIN_USER_TEMP_DIR per,
So the only difference is that Mac doesn't guarantee cleanup on
log out. (For example, if you're putting a Unix domain socket
in your runtime directory, when being portable to Mac you can't
rely on that socket being automatically deleted for you if the
user logs out but doesn't turn off their computer.)
ideally we use some (lib's) cross-platform approach for these
kinds of problems? I know in the past i've copied click.utils.get_app_dir()
impl for use, but not sure there's an equiv for all $XDG_ style
vars?
pretty sure @jaredgoldman had a fix for the posix shm
file-name len problem, but we may have lost that history?
Either way think a collab has generated a soln with ol claudy
Bo Fix: Shorten shm_list key for osx #341
Bc new users with fancy hw keep asking about it and i suppose once
we add support the project will prolly get popular with them "career
hackers" with "nice furniture" 😹
Immediately known (now solved) issues,
our use of
XDG_RUNTIME_DIRon
linuxin the UDS tpt backendfor the selection of the
UDSAddress.def_bindspaceis problematic.see this SO answer which
indicates using
DARWIN_USER_TEMP_DIRper,ideally we use some (lib's) cross-platform approach for these
kinds of problems? I know in the past i've copied
click.utils.get_app_dir()impl for use, but not sure there's an equiv for all
$XDG_stylevars?
platformdirsin (macOS support (for the fancy peeps with nice furniture) #406) B)SOLVED again via various workarounds in macOS support (for the fancy peeps with nice furniture) #406
macos..we now have Macos in ci #342 !!
Follow through after #406
there's our outstanding list from long ago ,
.ipc._shm._shorten_key_for_macos()may cause collisions for short names? #423 as a follow up tracking for symmetric shm-buf-keyscompared to the linux API.
we found out we need better
daemonfixture syncing supporttests.conftest: we need a betterdaemonsyncing approach, prolly via socket polling #424the test set should be better oriented around running
against multiple tpt backends, namely to start
'uds'--tpt-proto uds#422the final re-instatement of
windowssupport isn't veryfar off now either B)