The new haskell-ide repo.

Posted by Michael Snoyman - 26 October, 2015

Recently Alan Zimmerman announced on the haskell-cafe mailing list that there was a new haskell-ide project, with a new Github repository, a mailing list and an IRC channel. Some people have been concerned that this effort is fragmenting existing efforts, including with ide-backend (the open sourced library FP Complete announced earlier this year). I clarified this on Reddit, but wanted to take the opportunity to do so on this blog as well (and, additionally, throw some extra attention on the haskell-ide project).

Alan's announcement did not come in a vacuum; about two weeks ago, he reached out to others for feedback on a potential project. There were some side channel discussions that I was involved in, all of which were very much in favor of (and excited about!) this project. To quote myself from Reddit, we reached the following conclusion:

Both the ghc-mod and ide-backend maintainers have agreed to contribute code to this new repository and then rebase the old repos on this. The reason we're using a new repo instead of modifying one of the existing ones is so that the existing projects experience no disruption during this migration process. If this was a new set of people starting a new project without support from existing projects, I'd agree with you. But Alan's reached out to existing players already, which is an important distinction.

Michael Sloan - the current maintainer of ide-backend and one of the primary developers of both School of Haskell and FP Haskell Center - is already getting involved in this project. It's too early to decide exactly what the future of ide-backend will be relative to haskell-ide, but we're not ruling anything out. Anything from rebasing ide-backend to use haskell-ide internally, all the way to deprecating ide-backend in favor of haskell-ide, is on the table. We'll do whatever makes the most sense to help the Haskell community create great tooling.

Related to this project: a number of people have been following the development of stack-ide. We started that project not realizing how quickly existing tooling (like ghc-mod and hdevtools) would adopt support for Stack, and certainly not expecting this new haskell-ide project to offer a unifying force in the Haskell tooling space. To avoid fragmentation, we're currently holding off on further word on stack-ide, hoping instead that collaboration will help improve existing tooling not just for the Stack use case, but for cabal, cabal sandboxes, and other cases people have been targeting.

Since I'm already discussing IDE stuff, let me publicly give an answer I've given privately to a few people. A number of individuals have asked about the future of the FP Haskell Center codebase, and the possibility of open sourcing it. The summary answer is:

  • Everything related to School of Haskell is being open sourced. Most of that is done already, the rest is just in the last few stages of code cleanup.
  • The current state of the FP Haskell Center code base is difficult to simply give out, since it's based on some superseded technologies. For example, we created it in a world where Docker didn't exist, and have quite a few LXC scripts to make it work. We also have some difficult-to-manage build scripts that could be replaced by Stack. We've cleaned all of this up for School of Haskell, but have not done the same to the FP Haskell Center codebase.
  • As a general policy, we don't like to just drop unsupported code on the community at FP Complete. If there are maintainers that are interested in taking over the current FPHC codebase, we can discuss transferring it over. But simply open sourcing without any support tends not to be a helpful move (instead, it distracts from other, active projects, which we don't want to do).
  • One possibility going forward is that, once the School of Haskell web service is up, running, and stable, a new IDE project could be started that targets that same API. We're not planning on running such a project at FP Complete, but we'd be happy to provide some feedback, and include necessary changes to the SoH service to make it work.

I hope this excites others as much as it excites me: some concerted efforts on improving tooling can hopefully go a long way. A big thank you to Alan for coordinating this effort, and to Michael Sloan for leading the charge from the FP Complete side. I'm optimistic that we'll see some real strides forward in the near future.


Recent Posts

Big Data vs Business Intelligence: What’s the difference?

read more

DevOps Value: How to Measure the Success of DevOps

read more

Weakly Typed Haskell

read more