comment
This commit is contained in:
parent
7500ba7ceb
commit
ba7598dbab
1 changed files with 24 additions and 0 deletions
|
@ -0,0 +1,24 @@
|
||||||
|
[[!comment format=mdwn
|
||||||
|
username="joey"
|
||||||
|
subject="""comment 1"""
|
||||||
|
date="2021-05-12T17:09:06Z"
|
||||||
|
content="""
|
||||||
|
I'm seeing this too, although sometimes it seems to crash out before that
|
||||||
|
gets written to the log. And sometimes it manages to get running and seems
|
||||||
|
to run ok.
|
||||||
|
|
||||||
|
It does not happen with --foreground AFAICS.
|
||||||
|
|
||||||
|
Reproduced it with that most recent standalone release, and also with a
|
||||||
|
local dev build, also with ghc 8.8.4.
|
||||||
|
|
||||||
|
This would usually be a ghc or library bug, as there's no code in git-annex
|
||||||
|
low-level enough to cause this kind of memory corruption or whatever it is.
|
||||||
|
|
||||||
|
But, the assistant does one really funky thing for a haskell program, and that
|
||||||
|
is forking to the background. It seems just possible that doing that has
|
||||||
|
started confusing the garbage collector in some way.
|
||||||
|
|
||||||
|
forkProcess is really problimatic with its handling of threads etc,
|
||||||
|
and this may be the signal that git-annex needs to stop using it.
|
||||||
|
"""]]
|
Loading…
Add table
Reference in a new issue