ascheman commented on issue #1430:
URL: 
https://github.com/apache/tooling-trusted-releases/issues/1430#issuecomment-5037453129

   @sbp First, to place my perspective: I'm not (yet) a release manager — I'm a 
committer who joined the ATR alpha out of interest and hit this while verifying 
an RC, so take my "requirements" as an interested bystander's, not a PMC's 
release workflow.
   
   Thanks for the context (I hadn't seen #199 / the existing Compose-tab "RAT 
source excludes" field). For the immediate false positive that field already 
solves it, so no objection to closing this in favour of the existing mechanism 
if you prefer.
   
   On what I'd actually prefer, now that I've thought about it: the thing I'd 
most want to avoid is maintaining the same exclude list twice — and especially 
maintaining it through a UI rather than something derivable. A project already 
declares its RAT excludes once (pom apache-rat-plugin config, over the 
maven-parent defaults); they're reviewed, versioned in git, and travel with the 
source. Re-typing them in the ATR web UI per project is duplication that can 
drift from what the project's own `apache-rat:check` uses. So if the excludes 
can be *derived from the project*, I'd prefer that over any hand-maintained 
path — which again points at the atr-maven-plugin side (#78) doing the deriving.
   
   The one reason I can see *against* deriving-from-project is trust: excludes 
that come from (or ship inside) the release are controlled by the release 
itself, and a verification tool arguably shouldn't fully trust the artifact 
it's checking to tell it what to skip. If that's your concern, it's a fair one 
— and maybe the sweet spot is: the plugin emits the derived excludes as a 
*proposal*, and a committer still confirms/edits them in ATR (pre-filling your 
existing ignores instead of blank-slate typing). That kills the 
double-*authoring* without ATR blindly trusting the archive.
   
   To target #78 correctly, two questions:
   1. Which consumption path should the plugin feed — an in-archive file (exact 
path/name: `.rat-excludes` at root? `.atr/...`?), or your API/DB ignores (as a 
pre-fill)?
   2. Per-file globs enough, or do you also want a reason/scope per entry?
   
   I've no strong opinion on your internal storage (DB vs archive vs compose 
dir) — you know the cache-key/sync and trust implications far better. Happy to 
run a quick sweep across apache/maven-* to quantify how much projects rely on 
project-specific RAT excludes beyond your standard sets, if that helps judge 
whether the auto-emit is worth it.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to