Custom formatters examples/documentation?

[ccoVeille]

The README mentions this among the tool goals

Make it simple to extend with new custom formatters

The only formatter I found is the basic one available in formatter package

I looked for issues, code, discussions about custom formatters, but I found nothing.

Are custom formatters a thing? How to use them?

[braydonk]

The functionality to create your own formatter all exists, and I am currently working on implementing a new one that uses a new yaml backend. All the functionality to create new formatters exists, but I haven't actually documented any of it because there wasn't a ton of demand for it and most of the plans I had envisioned for it didn't really come to pass due to lack of time I'm allowed to work on the project.

There are no other formatters in the distributed yamlfmt binary, and I am not aware of anyone having created any custom formatters. It is technically possible to create a formatter that could work with using yamlfmt as a library, but since this is written in Go it wouldn't be possible to do something like distributing just a custom formatter that you load as a plugin at runtime. Any new formatters would have to be contributed to this repo to be part of the distributed tool.

I am currently working in my spare time on an alternate formatter that will use goccy/go-yaml, an alternate yaml library from the one that exists that fixes a few formatting issues the existing one has. It isn't ready for release yet due to some issues I'm having with it which I've opened issues for there. I've also considered alternate formatters for different languages like json or toml, technically possible I've just never had the time. I also am not convinced I could handle any increased issue load if I introduce something like a toml formatter and start to get new toml related issues, as I barely have time to handle the yaml ones I get already. However if someone wanted to contribute something like that and was willing to have issues assigned to them, I'd be happy to work with someone to get that figured out.

[ccoVeille]

Wouldn't it be more appropriate to wait for you to end your refactoring so instead of using this fork, it might be simpler to use the goccy/go-yaml

Because you plan to replace the lib by goccy/go-yaml write?

Or the goccy/go-yaml will replace the basic formatter available in yamllint ?

[braydonk]

That is another option. The new formatter that uses goccy/go-yaml will be made available as an alternate formatter called experimental, that will eventually move to a better name once enough people have tried it and provided feedback. (I'm also considering just naming it something more permanent and keeping both around for the foreseeable future, haven't thought about it enough).

I have a branch that works, but I don't want to launch it until at least goccy/go-yaml#614 is fixed, there could be other oddities I haven't dug into yet either but this is a pretty obvious one I want to figure out first.

[ccoVeille]

OK thanks again for the detailed explanations and context.

Basedon what you said, there is no real chance that the libyaml could be replaced.

It's more about an alternative formatter that you are working on.

So if I expect to have the formatting feature I talked about. Changes will have to be done in the libyaml, and later in your experimental formatter.

OK, now it's clear.

This discussion is really interesting. I'm unsure there was such a detailed explanations provided somewhere, with as many details.

Thanks again.

I think I will start opening a ticket about my need. Then I will see how much courage and time I can spend on trying to read, then update the your libyaml fork

[braydonk]

So if I expect to have the formatting feature I talked about. Changes will have to be done in the libyaml, and later in your experimental formatter.

I don't expect I will be able to delete the basic formatter within any foreseeable future; despite having not cut 1.0 of yamlfmt, since it's been around so long unchanging I'm hesitant to make any change that will break people at this point. So if you were to implement this in basic (i.e. in braydonk/yaml and then implemented as an option to the basic formatter) you can be pretty confident it will remain around. If you subsequently wanted to make the change to experimental as well you would be welcome to, though in that case I don't intend to fork since the maintainer is very active and responsive, so you'd likely need to work with that community closely to add any major features.

I think I will start opening a ticket about my need. Then I will see how much courage and time I can spend on trying to read, then update the your libyaml fork

Good idea, you may also be able to get community support from people who may be interested in the feature.

This discussion is really interesting. I'm unsure there was such a detailed explanations provided somewhere, with as many details.

Happy to help. Since I work on this project largely solo I end up doing a lot of private thinking and planning and not a lot of writing stuff down, so you can always feel free to open discussions if you have further questions.

[ccoVeille]

I opened

• feat: Format with padding #234