<domenkozar>
niksnut: can we include deadlock fix or is it too late?
<gchristensen>
domenkozar: it is nix 2.2 so should include it?
<gchristensen>
or is it unmerged?
<domenkozar>
unmerged
<gchristensen>
too late then, I think
<domenkozar>
:(
<niksnut>
domenkozar: as I commented, it isn't even clear whether using BSD locks fixes anything
<niksnut>
and it requires a schema change which is a headache
<domenkozar>
I guess we'll fix it the proper way :)
<domenkozar>
thanks!
<niksnut>
I mean, it would be great if somebody who can semi-reliably reproduce it could confirm whether BSD locks fix it
<samueldr>
hmm, I've been meaning to ask again since the past holidays wasn't active, but I hate the feeling of being "naggy" about it; if boehmgc's leak detector detects leaks within nix, could they be false positives?
<samueldr>
I haven't been able to make it print the position of the declarations of the leaks with nix, and do not know enough about boehmgc to really know how to check that
<niksnut>
Nix certainly leaks memory, in particular ASTs are leaked at the moment (since they don't use Boehm)
<samueldr>
right, how likely would it be that the leaks are the main issue with release-combined on hydra?
<samueldr>
talking about the heap things *searches the exact error message*
<niksnut>
not likely
<samueldr>
"Too many heap sections: Increase MAXHINCR or MAX_HEAP_SECTS"
<domenkozar>
niksnut: I'll try to reproduce it deterministically
<domenkozar>
haven't succeeded so far
pie_ has quit [Ping timeout: 260 seconds]
hedning has quit [Remote host closed the connection]
__Sander__ has quit [Quit: Konversation terminated!]