Question about Gitlab MR workflow.

Hello, I've just had a go at making a small MR on our new Gitlab system, and I am a bit confused about the intended workflow. I was following these instructions : https://gitlab.haskell.org/ghc/ghc/wikis/working-conventions/fixing-bugs My question is about the steps after 2 (i.e., 3 and 4).

Re: Old build system broken

Yeah so what I did is making settings no longer be created directly by configure, but by make and hadrian. I did this because I'm moving configurations options from Config.hs to there, Config.hs was generated by make and hadrian, and the whole thing will become stage-specific. I tried do

Re: Old build system broken

Some discussion on IRC with @Ericson2314 reveals that make maintainer-clean is not deleting settings files, which cause this bug. If you do a fresh checkout and build, the problem should go away. It's also possible that deleting inplace/lib/settings manually (and then running ./configure) may

Re: Old build system broken

Me too. I'm on a Mac. Deepest apologies (because I know this makes me useless), but I don't have the error message any more. It mentioned "Tables next to code" and the settings file, so I'm confident that it's related. Also, reverting 1aad97887747c351727ebd7b85217f2666f5b835 fixed the problem

Re: Old build system broken

Sorry to hijack the thread, I get something very similar on ppc64le linux: Configuring ghc-prim-0.6.1... ghc-cabal: '/tmpram/ghc/inplace/bin/ghc-stage1' exited with an error: No entry for "Tables next to code" in "/tmpram/ghc/inplace/lib/settings" libraries/ghc-prim/ghc.mk:4: recipe for target

Re: Old build system broken

That looks like stage1 has been improperly configured. Does /home/simonpj/code/HEAD/inplace/bin/ghc-stage1 --info *Return anything sensible for target arch? * *I still use the make system exclusively and haven't noticed a failure. * *But my nightlies haven't kicked off yet today. * *Thanks, *

Re: CI on forked projects: Darwin woes

I think there was the ghc-devops-group list, but I don't know if it is still active, and I kind of like to not have to follow too many lists. For example, I had also not realized that it is an option to push to branches on the main project, and have been using my own fork, so thanks for posting

CI on forked projects: Darwin woes

Over the past few days, I've submitted several merge requests from branches on my forked project (mostly because I didn't even realize pushing to a branch on the main project was an alternative). When those MRs run under CI, I've had a bunch of failures due to timeouts waiting on a

Re: Is cleaning up old issues worthwhile?

In my time working on an old system, I found the statement "they are so old that if the bug still exists there is probably a newer ticket" to be false multiple times.  Large systems hide many obscure bugs, some of which are rarely encountered. And sometimes the oldest symptom is the clearest

Old build system broken

I know we are supposed to be using Hadrian now, but is the old build system supposed to be broken? A clean build fails with "inplace/bin/ghc-cabal" check libraries/ghc-prim "inplace/bin/ghc-cabal" configure libraries/ghc-prim dist-install

Re: MR titles

Might I also suggest that we put a very brief summary of the area of GHC that the MR fixes in the title? With the way that GitLab puts numbers at the end of subject lines, it's harder to recognize tickets by number now. By including a few keywords in the MR title, I can find ones of interest

RE: MR titles

This is part of our contributing guide at work Do you have other advice from your contributing guide that we could learn from? Simon From: chessai . Sent: 08 May 2019 12:29 To: Simon Peyton Jones Cc: GHC developers Subject: Re: MR titles I agree that this is the way to go. This is part of

Re: MR titles

I agree that this is the way to go. This is part of our contributing guide at work, for the reasons you mentioned. Additionally it helps avoid the need to read the MR's contents (comments/code) before going into it, since reading relevant tickets is almost always something you want to do. Thanks

Parser performance: 10% regression in 8.8

Hello ghc-devs, This February I did some changes to the parser that require higher rank types support in ‘happy’. Unfortunately, as I discovered, happy’s --coerce option is severely broken in the presence of higher rank types, so I had to disable it. My benchmarks have shown a 10% slowdown

MR titles

Friends In this MR, Vlad includes the number of the ticket it fixes in the title of the MR. That is so helpful: * It links the MR back to the issue * And does so in large font (screen shot below) * And the link is clickable, even