Sorry for the late response.

I just returned from vacation and joined the mailing list.


This is a response to the last comments from the JIRA ticket:
https://issues.apache.org/jira/browse/RAT-337


*1.) Rather than pulling the repo in (which would, most likely, involve the
Incubator (in other words: All kinds of bureaucratic processes), I suggest
that we simply copy the files action.yml, json.xsl, and possibly README.md.*


I think transferring a repo should not be that complicated.

There are only four files. I already appended the *Apache-2.0* licenses
headers to each file.

The only file missing, if needed, is the NOTICE file.


Technically, we could create a new repository using the *ASF Self-serve. *I
could then submit a PR with these files.


I believe we can retain the commit history by merging without squashing.


*2.) The main question for me is the target: Could we simply import these
files into our main repository, or do we really need a separate repository
within Github?*


Maybe the content could be imported to the main repository, but you would
need to figure out how you want to manage the tags.


All other Actions use the following tag naming convention.


v1

v1.0.0

v1.0.1

v2

v2.0.0

v2.1.0


E.g.

- https://github.com/actions/setup-node

- https://github.com/microsoft/setup-msbuild

Basically *v1* and *v2 *will represent the latest version of that major
release. With ever new minor or patch version, we force update the v<MAJOR>
tag with the latest commit id.

This allows peoples to fetch the latest major version automatically without
updating their workflow.


Looking at the current tags in *apache/creadur-rat* repo, I am not sure if
introducing these tags will confuse anyone, so I feel a new repo could be
better?


I never saw an action tag starting at v0 but maybe its not an issue. Then
it could be inline with rat's release versions.


v0

v0.15.0

The *action.yml* file would have to live in the *root *directory. The other
files could be in its own folder but we would have to update the yml file
appropriately.


*If the latter, we could probably ask Infra for a second repository within
the ASF, with an associated mirror on Github,*


For the Apache Cordova project, which I am a part of, all of our ASF repos
are mirrored on GitHub. I thought this was a default setting, but I could
be wrong.

Last time I created a new repo with the *ASF Self-serve,* I believe it was
mirrored to GitHub for us.


I* am leaning more towards a separate repo* because I don't think it's
necessary for CI run to checkout the entire Apache Rat repo on every run.

Depending on how the GitHub Action workflow is configured, it could run on
every commit and to checkout the entire repo on every commit could be
excessive. Though I do not believe the repo is that large.

Another reason prefered on GitHub is because I never saw or tried to import
from an external repo service. It might be possible with a full git URL
path.

Reply via email to