No Maintainer

by @fommil
15th September 2017

In December, when my first child is due, I will give up my online hobbies to be with my family, and finish my book, Functional Programming for Mortals with Scalaz. Of most relevance to ENSIME is that I will, sadly, no longer have the time to maintain this project.

I have anticipated this over the last couple of years and automated as much as possible (e.g. the server, Emacs, docker and website release processes) to make it easy for any motivated user to contribute features, bug fixes, and code reviews.

My real hope was that the sponsorship programme would generate enough revenue to pay somebody to take this role. But even with the exceptionally generous donations from our core userbase, our funds are orders of magnitude lower than necessary.

ENSIME needs people who are able to do the basic job of assigning authority roles (such as merge rights, publishing to Sonatype), cutting stable releases of the various components, keeping the servers ticking and taking responsibility for code / build quality (i.e. code reviews and merging).

The ENSIME contributor base is absolutely fantastic and it is a genuine pleasure to hack with the great people who come through these revolving doors. There is an energy to being closely involved with this critical part of the scala ecosystem, whilst being a banner for Free (libre) Software, empowering power users to hack their workflow to their exact requirements.

It is addictive to contribute to a project where you are the customer, not somebody else. Unlike at your day job, where projects can live or die by the quality of communication between the customer and developers, with ENSIME you know exactly what you want to do and you are in the unique position to be excited about the possibilities of what can be done AND the implementation details.

That said, there are a lot of additional burdens that you, as the future maintainer, will have thrust upon you unless you actively identify the lures and choose not to get involved.

In the interests of being honest, this is some of what you can expect as a Free Software maintainer:

  1. keeping a mental picture of the overall project: server, all build tool plugins and all editor plugins. You will be expected to have a plan for what the next milestone looks like and be aware that any pull request you accept can break somebody’s workflow. Like the maintainer before you, tests will be your best friend. I actually quite liked this architectural bit. I’ve created a plan for the 3.0 series, feel free to do what you want with it.
  2. being expected to review every pull request and give advice on implementing features that you’re not really excited about. Sometimes this might mean finishing off the PR yourself because the contributor is stuck. Be harder than me, just park the PR.
  3. deciding your stance on backwards compatibility and how much effort you will spend on ensuring that you don’t break the workflow of somebody who is not contributing to the project. Before making any large breaking changes, we have in the past cut a stable release so that nobody is left without the ability to use ENSIME. This is a burden to you, but it will save you from having to deal with angry users.
  4. being aware of upstream developments in scalac, dotty and sbt, and getting involved in the scoping of those projects to ensure that they don’t result in irretrievable situations where ENSIME is no longer viable. e.g. writing requirement documents for those developers.
  5. continually upgrading ENSIME to support the latest scalac, dotty and sbt even if these things are not part of your daily usage. Although the community is good at the occasional bugfix or feature, the larger issues like this require a larger commitment. ensime-server was at one point part of the scala Community Build, but it became too burdensome to maintain. Be aware that nobody upstream checks for ENSIME compatibility before releasing and you will be expected to fix it if it breaks.
  6. telling contributors that they cannot add any more dependencies because it makes upgrading to newer versions of scala and sbt much harder. Remember that you will pay the cost if you listen to them.
  7. keeping the documentation up to date on the website. Be prepared for lots of passive aggressive insults if there is even a single sentence that is wrong or it didn’t make sense to somebody who just skims it. Don’t be afraid to point out the [edit] button that I added to every single page.
  8. there is a minority of random dudes on the internet who believe you owe them something: they believe themselves to be The Customer and are therefore Always Right. Be prepared for the expectation that you are available 24/7 to spoon feed your documentation to them, and that you shouldn’t close their “bug reports” even though its covered in the Troubleshooting Guide and they ignored the template (which asks for information that you absolutely need to help them). If you don’t decorate your advice with salutations and graciously thank them for trying out ENSIME, be prepared to be called arrogant, lazy and rude. They seem to expect to be treated like a coffee shop customer and genuinely believe they are helping you by using your software. My advice is to continue making it clear that ENSIME is about empowering hackers who wish to control their tools, and shut these “customers” out of the project if they turn nasty and not to tolerate them, but you will be ostracised by other groups in the Scala ecosystem who wish to encourage privileged behaviour. They will call you toxic. You will need thick skin to deal with both of these types of people.
  9. having a Code of Conduct. If you do not have one, you will be pressured by people until you put one in place. Their primary weapon is emotional blackmail, they will claim that you are excluding people until you put their code of conduct in place. They will exclude you from their social gang / old-boys-club until you agree with them. You’re between a rock and a hard place: no matter what you do, there will always be somebody who complains about it. For example, it seems we need to have a stance on “actual Nazis”, and you need to prove your commitment to the cause by punching a Nazi, or something. This is why we have a simple code of conduct which asserts that users should be respectful of the contributors’ time and help each other out. ENSIME is a technical project and the only politics we subscribe to are those of the Free Software Foundation.
  10. managing the sponsorship programme and the cash funds available there. We are currently generating about $350 / month, which means we can afford to hire a part time student for two months a year. By December, the funds will be depleted. It’ll be your responsibility to basically do all the management and admin around this, including finding the person in the first place (do you have any idea how hard it is to find good Scala developers?). The effort involved in doing this is quite high, although the payback is good. You may want to consider canning the sponsorship programme because of the admin burden, and give what little remains to the Free Software Foundation.
  11. The cost of appveyor (Windows CI) is $300 / year. Your choice of CI and worker agents varies, currently running at about $350 / year plus Chip’s machine (which does the lion’s share of the work). Stickers (everybody loves stickers!) cost around $100 a batch. You can take money out of the sponsorship pot for this, I didn’t. You may be expected to top up the sponsorship funds if they are running low and the developer needs to be paid because of timing. I recommend canning the Windows CI and letting a Windows user contribute the funds and effort to keep the Windows build running, unless the gratis AppVeyor service has improved in the last two years (which you will need to spend a weekend investigating). I moved us over entirely to Chip’s machine for GNU / Linux CI.

Applicants need not apply, I will grant admin permissions to some core contributors over the coming months and somebody will eventually step up out of necessity. There will be no handover, you will (re)discover and (re)invent everything as you need it, much as the incumbent earned his scars.

Frankly, I wouldn’t wish the burden of Free Software maintenance on anybody. It’s too much abuse, and the politics are painful. We are all the maintainers now: