cstamas.bsky.social
Resurrecting dead software, Apache Maven and alike.
80 posts
240 followers
294 following
Regular Contributor
Active Commenter
comment in response to
post
Checkout github.com/apache/maven... (but is scheduled for 4.1)
comment in response to
post
Simplest way to use it explained here: maveniverse.eu/docs/njord/u...
Just install the extension, and `mvn deploy -DaltDeploymentRepository=id::njord:release-sca` or alike
comment in response to
post
Njord is not invasive, on contrary, you can use it without changing anything
comment in response to
post
Give Maveniverse Njord a try: maveniverse.eu/docs/njord/
comment in response to
post
Site is still to be done properly. Whole thing is still in phase when source is the main documentation :)
comment in response to
post
Now that could go into the core, as the core knows what mandatory config is for given mojo...
comment in response to
post
Well, this is explained in POM reference maven.apache.org/pom.html#Plu... where you can apply config "to plugin (as whole)" and "to specific execution". What user did is completely valid, but probably not what he wanted, but XML structure nicely reflects the "scope" of config.
comment in response to
post
Right. Perso am not delighted as both sonatype plugins share the same dna, and history for them is not the greatest...
comment in response to
post
How would you control what to end up there?
comment in response to
post
This check was present since long time in Takari Lifecycle and was one of the features I LOVED about it takari.io/book/40-life...
comment in response to
post
@lobaorn.bsky.social and @sormuras.bsky.social helped me figure out that an argument file could bypass the Windows limitation.
Both Maven and the IDE seem happy so far 🙂
comment in response to
post
In another (related) direction, all major projects focus on Java 17+, with the same growing pains, and it is nice to see so many issues linking each other. One new issue that I saw this morning was from Netty, from @cstamas.bsky.social regarding using Maven 4: github.com/netty/netty/...
comment in response to
post
You are aware, that tool runtime != targeted plaf? Plus, mvn4 has improved tool chains support like auto discovery...
comment in response to
post
Did you mean .mvn/settings.xml?
comment in response to
post
There is still the "local repo hygiene"... so you get a coffee when you nuke your local repository? As this is the main goal: when you throw away you local repo, w/ mimir it is "still there" and builds are still fast: cstamas.org/blog/2025/01...
comment in response to
post
I have a desktop (Linux) and two laptops (Linux and macOS) I use interchangeably, and I _hate_ when I move from one to another to "pick up" all the deps" of project am currently in. With Mimir, I get "shared" cache with laptop on LAN speeds and is almost as fast as doing git update and just continue
comment in response to
post
This basically helps you to simply nuke your local repo, even daily, but NOT lose anything (time) with "downloading the internet", as it is already there (mimir today caches Central ONLY, will be improving on that front).
comment in response to
post
w/ system-wide cache, you have ONE artifact copy on your system as rest are hardlinked from Mimir local cache. Moreover, "nuking local repo" is part of (currently) of repo "hygiene" exactly due it's mixed nature, and is especially painful if you work on multiple branches of several inter-dep prjs
comment in response to
post
So far, you'd end up with as many copies of artifacts as many local repositories you have (I like to keep things separate). Second, in local repository remote artifacts are _mixed_ with locally built ones, and may be even replaced (ie. you check out a project tag and do mvn install w/ soma change)
comment in response to
post
Same here: cookie free and fully static. Simple.
comment in response to
post
Just tested Mimir w/macOS (so far was playing with Linux desktop and laptop) and works just nice (either as consumer or as publisher). So again, as long as we talk about Linux or macOS the NIO2 is the best choice (and works as one would expect, and atomic ops are atomic)... on Win, all bets are off
comment in response to
post
All in all, out of 3 OS we test with, two FS behave "as one could expect", while one FS is "like from other planet". We have promised access to s390x GH runners (MNG-8284) and I am very curious how this fourth OS will behave...