golines will probably be archived in the coming months

[ldez]

golines will probably be archived in the coming months (it's written in the readme).

As of late 2024, segmentio/golines has functionally been in maintenance mode and several dependencies appear to be similarly unmaintained.
At some point in Q4 2025, this repository will be archived unless active maintainership can be found within Twilio Segment.
The code will remain available, and the terms of the license will not be changed.

Based on the text inside the readme, I think they don't want to transfer it.

So I started working on a fork that's more complete than the one we currently use.
I used it to refactor the code, set up more complete tests, and fix some bugs.

https://github.com/golangci/golines/tree/wip/main

I'm still working on the branch (and I push force it), but the overall state is stable and without breaking changes.

I'll open an issue on the original repo for discussion with Segment, but I'll wait a bit in Q4.

[ldez]

I'm still working on it, but I think I've made good progress.

Still, nothing is breaking.

My primary goal is to enhance the project's maintainability, as this type of project can be challenging to maintain.

Project Structure

I reorganized the project with packages.
First, because we need it for golangci-lint.
Second, because it will be easier to maintain.

I replaced logrus with slog (the logger is only used for debugging).

I still haven't decided if I will convert the shortener package to a module (to avoid some dependencies when using it as a lib)

Tests

I added a new test system that allows testing all the Shortener options.
And so I added more tests.

I also added more tests on each structure, function, etc.

CLI Performance

I improved the performance of the CLI significantly by using the same design as go fmt, and some optimizations.

My fork is about four to six times faster than the current golines.

Note: I don't have a recent laptop (~4 years old).

On the Kubernetes codebase:

		Time
current	goimports as CLI	4m45s
fork	goimports as CLI	51s
fork	"goimports" as lib	46s

		Time
current	"gofmt" as lib	1m40s
fork	gofmt as CLI	29s
fork	"gofmt" as lib	25s

It was not possible to use gofmt as CLI with the current version.

Skip cases

I added more skip cases (testdata, dot directories, etc), basically the same thing as golangci-lint fmt.

Bugs

I fixed some bugs:

• wrong directive pattern
• struct tag literals
• etc.

Notes

The default base formatter (goimports) didn't provide a good experience by default.
It was four times slower than gofmt.