Subscribe to this issue to get requests for user feedback!

[casey]

Please subscribe to this issue if you'd like to receive notifications for requests for user feedback.

Situations often arise in which user feedback would be very helpful. Some examples are:

• Questions about whether a potential new feature is worth it or not.
• Questions about how a new feature should be implemented.
• Issues concerning existing features that I don't use.
• Issues concerning Windows, since I use Unix machines personally, and don't have a lot of Windows dev experience.
• Weird edge-cases where I can't figure out whether something is a bug and should be fixed, even if it might break existing justfiles.

If you're interested in helping out, subscribe to this issue! This issue is locked, so the only notifications that you'll get will be requests for help from me. I'll always link to another issue, and keep commentary out of this thread, so it should be focused.

[casey]

First request: Check out #1201 for a proposal for a backwards-compatible way of introducing otherwise breaking changes. Tell me what you think. Is this a good idea? Is there a better way to do it? Are there any pitfalls I haven't thought of? Will it be super annoying if people have to put set edition := 2 to get better defaults?

[casey]

Second request: Initially, just used sh on all platforms. This required installing sh on Windows, but made justfiles work the same everywhere. Later, the set windows-powershell setting was introduced, which allowed using powershell.exe on Windows but sh everywhere else. PowerShell and sh are often compatible, so this was a good way of writing cross-platform justfiles. Later, Microsoft introduced a new version of PowerShell as pwsh.exe. The set windows-powershell setting can't be changed to use pwsh.exe, because the two version are incompatible, and some justfiles using set windows-powershell might break. So, it is now proposed that we adde a new setting set windows-shell := […] that allows setting an arbitrary shell for Windows. This allows users to opt-in to whatever shell they like on Windows, and is probably what we should have done from the beginning. I'm pretty sure this is a good change, but I don't use Windows, so I'm always nervous about things like this. Comment and let me know what you think! See the PR here #1198 and the issue with comments here #1190.

[casey]

Third request: A proposal for a new recipe type, which solves a number of issues and inconsistencies, and could be made the default recipe type in a future edition.

[casey]

Fourth request! Just now has modules and imports.

Imports load a source file and include all items in the source file in the current justfile.

If a justfile contains:

import 'foo.just'

And foo.just contains:

foo:
  @echo FOO

Then running just foo will print FOO.

Modules create independent submodules, whose recipes can be invoked as subcommands:

If a justfile contains:

mod foo

And foo.just contains:

bar:
  @echo BAR

Then running just foo bar will print BAR.

Imports are stable, but modules are currently unstable, so you'll need the --unstable flag to use them.

Modules are a big feature, so I'm hoping to get plenty of testing and feedback before stabilizing them. Try them out and let me know what you think! You can see planned features and leave feedback in #929.

[casey]

I want to give a final call for testing and feedback for the modules feature. All bugs I know of have been fixed, and I added a source_file() and source_directory() for getting the file of the current source file, and the directory that contains it, respectively.

I think it's pretty much ready to go, and I'll try to stabilize them in a month, assuming nothing goes wrong. Check out the tracking issue to leave your comments!

Also, while I have your attention, we just released 1.27.0, which has a ton of new features and functions, so check that out too!

[casey]

I've landed a few PRs that improve --fmt output recently:

• Preserve comments before [doc] attribute #3275
• Reify newlines as AST items #3274
• Omit default boolean values when formatting #3268
• Preserve trailing comments when formatting #3282

I now have a bit of a dilemma. I'm starting to think that formatting output is good enough that the feature could be stabilized. However, I think that formatting is a feature which is likely to be subject to tweaking over the long term. Better ways to format the same thing (like breaking long expressions onto multiple lines), and fixing random bugs.

Given all that, there are a few options:

1. --fmt is permanently unstable
2. --fmt is stable but can change in a minor just release and isn't bound by just's normal stability guarantee (y in x.y.z)
3. --fmt is stable but we can't ever improve it

I think that 2. is probably the best option, but it's a departure from the usual extremely strong stability guarantees. If we did 2., I would probably still try to avoid frequent changes, which, most annoyingly would break CI jobs that use --fmt --check, maybe keeping them to once or twice per year.

Let me know what you think in #868!