Closed felixwellen closed 6 months ago
I think we should stick to removing the line RTS_OPTIONS=+RTS -H6G -RTS
, since it causes problems with newer ghc versions (than we are using in the CI - that's why it didn't show here).
Can someone (@phijor @jpoiret ?) check the flake and do nix flake update
and commit the result?
I've updated the flake inputs locally; I'll get back to you with the result once it's finished building.
NB: I'm not a contributor to the repo, so I cannot directly push to this branch. I can put the commit in a fork of mine though.
Thanks!
NB: I'm not a contributor to the repo, so I cannot directly push to this branch. I can put the commit in a fork of mine though.
Yes, just do that and I'll drag it over here.
I don't use/have Nix, so I can't really help with flakes, but I'm doing my best keeping it updated on Guix!
Looks all good to me now!
@mortberg @mzeuner @ecavallo are you okay with a release or should I wait for anything with merging this? If you have the time, maybe it is good to test it on your machine, since there was some trouble with the make file - I noticed some weird behaviour around RTS options with newer GHC versions.
I actually missed one instance of "Experiments"-exclusion in the makefile - maybe this went by the CI because of caching...
It is fixed, looks all good to me now (I checked 977 files in 25 min, so there is also no tc-time blow up).
I assume no complaints means I can go ahead with merging this and doing all the release stuff (agda 2.6.4.2 is around the corner..).
On one of my machines, there is somehow a too low default maximum for the heap. Setting a high maximum (I chose 32G) solves that and I don't see any harm in that, so I'll leave that in as it is now.
The suggested heap size also seems to set a maximum. I don't why we have a suggested heap size anyway, so I just removed that, since make failed with agda 2.6.4.1, compiled with ghc 9.8.1.