Can you add some notes regarding this cross-compilation configuration to the wiki?
I just tried to make FAST=YES in libraries/base and found that the makefile we used to have to support this is no longer there. This diff seems to be the culprit. I can work around it for now, but FYI this was useful!
This will do.
Tue, Feb 20
@angerman, I hope you don't mind if I revert this for the moment so we can try again to commit a cleaner version and work out the Hadrian backporting details.
Also, why does libraries/libiserv/src/Main.hs exist?
See my comments on D4377.
Hmm, this is rather problematic from the perspective of the stable branch as it requires changes to Hadrian. It looks like we will either need to cut a ghc-8.4 branch for Hadrian (and backport commits from master as appropriate) or cherry-pick this patch to the GHC ghc-8.4 branch to avoid divergence between stable and master.
Indeed, these timestamps have annoyed me in the past as well.
Mon, Feb 19
I need to have another look at this.
In this build I just so break the upper max_bytes_used limit for T1969: 15.9% on a 15% limit.
Maybe it would be better if I'd move the most-common-RHS code into -O2!?
Indeed I would be rather surprised if these weren't bitrotted.
But do let me know if I've misunderstood; the last comment on D4394 is a bit ambiguous.
What is the status of this, @tdammers?
Alright, it looks like this should be closed in favor of D4394.
Looks good to me!
Based on the coding conventions, shouldn't we rename the extension LangExt.GeneralisedNewtypeDeriving?
As a side note, I don't know where we guarantee that the string literals get floated to the very top. I think that's the only kind of floating that the simplifier does, but I couldn't find any comment on that.
Thank you for doing this, @niteria!
Sun, Feb 18
Yay! That being said, I'm really not pleased with the repetition here. I understand we can't use symlinks. Perhaps we could instead teach the build system to copy the sources into place?
I suppose this is reasonable. However, doesn't this mean that we will oversubscribe users running CircleCI on their own forks that lack the higher resources limits that we enjoy?