Add missing ASF license

[pingtimeout]

Note for reviewers: the ASF guidelines says that "LICENSE and NOTICE files belong at the top level of the source tree". But I think it would be akward if there was a LICENCE file at the root of the polaris-tools repository but not in any of the tools that are packaged independently.

[jbonofre]

As I guess it's for source distribution, it can go at the root level.

[flyrain]

I agreed with @jbonofre that one license file is good for src distribution, but we will need separate ones if we want to distribute them separately, like a binary distribution for the migrator.

[pingtimeout]

@flyrain should we revisit this when we start releasing the tools?

[jbonofre]

I think this LICENSE is incomplete (assuming we plan a source distribution for polaris-tools):

• it should mention gradle wrapper present in the subfolders
• it should mention that catalog migrator is coming from Nessie

Also the NOTICE is missing in this regard.

[pingtimeout]

The NOTICE file will come with a separate PR, as I am still figuring out what should be in there. Regarding the LICENSE file, do you mean that it should be the union of all the LICENSE files that are in subdirectories? If that's the case, I am in favor of not having such a file at the root of the repository as the value is not clear.

[jbonofre]

No I mean we would need two kinds of license: one for source distribution, one for binary distribution. The content often differs.

[jbonofre]

And also yes the root license should be valid for all subfolders (from source distribution standpoint)

[pingtimeout]

Now I am even more confused :D

AFAICT, the current LICENSE file that exists in the catalog migrator is not valid, as it does neither include the license of the plugins nor of the dependencies. So I am not sure whether it would be for the source or for the binary distribution.

I suspect that this source vs. distribution distinction is why we have a LICENSE file at the root of the polaris repository as well as one at runtime/distribution/LICENSE.

Could you confirm that this assumption is correct? As far as I can tell, https://infra.apache.org/licensing-howto.html only applies to bundled dependencies, i.e. for the binary distribution LICENSE (and NOTICE).

[jbonofre]

Not only. It's also for code copy

[jbonofre]

Not only. It's also for code copy. Let me update this pr you will see.

[snazy]

The idea is AFAIR that each tool is independent from the other tools.
Each tool has its own release cadence, if any.

Tools that get released would produce a source-tarball with the contents of the tool-directory,
not the whole repository.

I think we need two "levels" (cannot come up with a better term) here:

• "Per tool" "top level" LICENSE and NOTICE files - these contain the tool-specific license/notice information. As a Podling, I think the DISCLAIMER should be copied there as well.
• "Generic" or "umbrella" LICENSE and NOTICE files in the root folder that reference the tool-specific ones. These files are needed for the `git clone" or "git archive" cases.

If that approach works (@jbonofre may confirm or not?), we'd need LICENSE+NOTICE for the benchmarks + polaris-synchronizer tools.

[flyrain]

Yes, we could revisit once we are releasing the binary package.