Closed andreasabel closed 1 year ago
@Ericson2314 : What do you think? Would you be willing and able to do this soon?
It would help the Haskell community to prepare for GHC 9.6. E.g. testing BNFC on 9.6 is currently stuck on happy
.
@Ericson2314 : I can also do it myself if you add me on Hackage: https://hackage.haskell.org/package/happy/maintainers/edit
I can also do it myself if you add me on Hackage
Thank you. I added you to the maintainers.
Oh I am sorry I missed this.
@Ericson2314 started work on this, so I won't interfere:
@andreasabel I am not sure why that CI is failing but I also think it's fine if we make the release before merging it back into master.
I went to bed not having done that yet, and to make up for my tardiness I'll gladly do it myself today, but I don't want you to think me being slow => me wanting to do the release myself. You should certainly be fully empowered to not have to wait for me! (And hopefully now you have all the perms so that is indeed the case.)
To test my new capabilities I uploaded a candidate: https://hackage.haskell.org/package/happy-1.20.1/candidate
And published: https://hackage.haskell.org/package/happy-1.20.1 ! The candidate tested well when I downloaded and installed it, and the changelog looked good.
I pushed the tag 1.20.1
to the current HEAD of release-1.20
.
@Ericson2314 : Thanks for fixing the changelog and the .cabal file!
@andreasabel thank you for doing the main work and keep on top of things!
So I botched this job...
Here is a new candidate: https://hackage.haskell.org/package/happy-1.20.1.1/candidate
This was generated with make sdist
rather then just cabal sdist
.
The candidate tested well when I downloaded and installed it,
I guess my tests were not stringent enough, since I have a happy
in my PATH on my system. Thus, bootstrapping was not tested.
@andreasabel Maybe one of us should take a crack at doing github actions on the release-1.20
branch too. I fear this won't be the last time we release a new patch release of it.
Let me give it a try.
Thank you!
After trying a patched Haskell-CI first, I wrote a CI manually and pushed it to the release-1.20
branch: https://github.com/haskell/happy/actions/runs/4118004445
It basically does make sdist && cabal build
(and runs the tests) on a fresh system with a ubuntu-released happy
preinstalled.
It could be improved by saving the generated tarball as artifact and then installing happy
from this tarball in a subsequent job on a fresh system. This way, we could verify bootstrapping.
It could be improved by saving the generated tarball as artifact and then installing
happy
from this tarball in a subsequent job on a fresh system.
I implemented this and pushed: https://github.com/haskell/happy/actions/runs/4118983519
The CI now implicitly explains how to distribute happy (if you read it carefully) and verifies it for GHC versions 8 and 9. (GHC 7 bindists aren't reliably available on recent Ubuntu.)
Ok, hope this is done now, happy-1.20.1.1
is out. I pushed the tag 1.20.1.1 to the current HEAD of release-1.20.
Note to myself on how to release happy-1.20
:
make sdist-test
;-)
Thanks so much, @andreasabel!
See https://github.com/haskell/happy/compare/1.20.0...release-1.20
This will fix the
mtl-2.3
incompatibility issue ofhappy-1.20.0
:236
Work pushed to
release-1.20
, which I branched off tag1.20.0
:Control.Monad
1.20.1
, updated maintainer andtested-with
ANNOUNCE
,TODO
,README.md
from package@Ericson2314 , @int-index , @simonmar : I do not have the Hackage rights to release myself, so please
I tested building with GHC 7.10 - 9.6.0 locally. The code changes are minimal (
import
fixes in one file), so I do not expect trouble.