LUCENE-10234: Change module prefix to org.apache.*

[dweiss]

No description provided.

[uschindler]

I added a change to the CHANGES.txt file to explain that the automatic module names are a preparation for full module system support. This should be added, because we have not fully tested that everything works well with automatic module names.

Also I added a note that the module names should not be considered "stable". OK?

[rmuir]

looks good to me

[mocobeta]

Also I added a note that the module names should not be considered "stable".

For what it's worth, I don't think this notice is needed. I've seen many small/large projects that have changed artifact names for various reasons; users have dealt with it (I don't much care about the artifacts/modules name is stable or not, I just search for the right artifacts I need on the Maven central and it works). While the frequent naming changes would be confusing, do we need to strictly define stable names somewhere in the future...? - It's not an opinion but my feeling.

[mocobeta]

I checked luke.sh and luke.cmd works for me, just in case.
+1.

From my experiences, launch scripts are the ones of the most fragile and often overlooked parts in software projects. I think Windows bundles PowerShell for years which supports POSIX shell scripts, could we discard .cmd somewhere in the near future releases...

[dweiss]

@mocobeta the powershell scripts have much stricter isolation/ security checks and they would have to be signed. I'm not sure how this works but my experience on github actions infrastructure is that if you start messing with powershell, things stop working.

https://docs.microsoft.com/en-us/powershell/module/microsoft.powershell.security/set-executionpolicy?view=powershell-7.2

[dweiss]

"Pull request authors can't approve their own pull requests".

...but I do.

[dweiss]

I'll move this snippet to the dedicated jms section of the build (in the PR that adds full module support). It is useful.

[uschindler]

If we have full module support, the automatic module name attribute in manifest won't be there anymore. We may need other code to display this.

[dweiss]

Yes, I know. It still belongs over there though.

[mocobeta]

the powershell scripts have much stricter isolation/ security checks and they would have to be signed. I'm not sure how this works but my experience on github actions infrastructure is that if you start messing with powershell, things stop working.

ah, okay... anyway thanks for maintaining it correctly.
I don't think we want to repeat the manual tests for the luke start scripts every release or receive bug reports from users right after publishing the artifacts. If we need them for future releases to come, I'd like to seek a way to integrate a sanity check for it into our smoke test or CI.

[dweiss]

I'd like to seek a way to integrate a sanity check for it into our smoke test or CI.

I know how to do it - I am doing it elsewhere. I just need to get to this part. Track that jms pr, please!

[dweiss]

@uschindler Let me know when you think it's ready and I'll merge/ backport.

[uschindler]

I don't think we want to repeat the manual tests for the luke start scripts every release or receive bug reports from users right after publishing the artifacts. If we need them for future releases to come, I'd like to seek a way to integrate a sanity check for it into our smoke test or CI.

This was also always an issue with Solr, which stopped working on Windows or Linux from time to time. Solr's Smoketester checks this (although not the whitespaces), but Solr does not need a GUI. For Luke it is harder, because we can't easily check a GUI. One idea would be to add some command line parameter to Luke's main class like --sanity-check. If you pass this to script it is justpassing to main function which does some tests if all works.

[mocobeta]

I know how to do it - I am doing it elsewhere. I just need to get to this part. Track that jms pr, please!

One idea would be to add some command line parameter to Luke's main class like --sanity-check. If you pass this to script it is justpassing to main function which does some tests if all works.

Thanks, I would not like to do further thread hijacking here, but I'd love to continue the conversation on another issue or PR.

[uschindler]

@uschindler Let me know when you think it's ready and I'll merge/ backport.

Let's wait till this evening to get everybody a chance to bring in hisher opinion. Christian Stein from OpenJDK/JUnit/Maven already answered on the issue.

I sent a HEADS UP as suggested by Adrien to the Mailing list yesterday evening.

[uschindler]

There was positive review, also from Elasticsearch people, on the dev mailing list, so I think we can merge this. Maybe later this evening or tomorrow morning?

So my +1 to merge soon.