niri: update to 25.01

[trolljoe]

Testing the changes

• I tested the changes in this PR: NO

this goes by the new packaging niri page (https://github.com/YaLTeR/niri/wiki/Packaging-niri), if setting XDG_RUNTIME_DIR to mktemp isn't enough i'll skip the tests.

[trolljoe]

tested and it seems to work fine.

[lukeflo]

tested and it seems to work fine.

Can confirm that it works just fine!

[oliver-cfc]

Another confirmation that all is good... on a musl installation here.

[chilledfrogs]

Works fine for me!
Might I ask, however, why is the dbus feature not enabled?

[classabbyamp]

would be better to use something like v$version, as

1. hardcoding things like that is bad and will be forgotten at some point
2. we build from the tag, which is a reference to the commit

[chilledfrogs]

This variable is in accordance with Niri's own packaging guidelines, it is on top of the reported version number, i.e. niri 25.01 (e35c630)

[classabbyamp]

upstreams do stuff like this all the time. if they aren't leaving the commit in their release tarballs somehow (like a file), we shouldn't need to hard-code it. we aren't building some random commit, we're building the release, so the commit information is redundant.

[chilledfrogs]

Understood and agreed that it is redundant, however if it is expected by upstream, I'm not sure I really think it is worth it to break that expectation downstream is all (i.e. for bug reports and such)

[classabbyamp]

If upstream wants this information to be accurate, they should add it to their release artifacts. they already generate them to vendor cargo dependencies.

[chilledfrogs]

Fair point, I might poke upstream to add that, but then tbh I think it is best to just not add any environment variable as @trolljoe did now, since it will report (unknown commit) by default, which is more meaningful here

[classabbyamp]

maybe:

pre_check() {
    export XDG_RUNTIME_DIR="$mktemp -d "$wrksrc/runtime.XXXXXX")"
}

this will keep things contained inside the build directory and make it only used for tests

[trolljoe]

is the syntax correct? wouldn't i use this instead?

pre_check() {
    export XDG_RUNTIME_DIR=$(mktemp -d "$wrksrc/runtime.XXXXXX")
}

[chilledfrogs]

Methinks it is a typo, but mind the additional quotes:

pre_check() {
    export XDG_RUNTIME_DIR="$(mktemp -d "$wrksrc/runtime.XXXXXX")"
}

[classabbyamp]

yes mine was missing a (

[trolljoe]

Works fine for me! Might I ask, however, why is the dbus feature not enabled?

i will update the template including that, my bad

[chilledfrogs]

i will update the template including that, my bad

Np, there may be legitimate reasons for it actually since I am not 100% sure what it does (but Works For Me™ without elogind for instance), but I feel it may be useful, so I wouldn't mind if someone more experienced than me on these matters could weigh in too

[lukeflo]

Might I ask, however, why is the dbus feature not enabled?

@trolljoe it wasn't enabled in the former version, if I'm right? However, I didn't have recognised any drawbacks even without dbus feature, simply starting niri with dbus-session-run niri. Just as information... 😉