[poll] need opinions on ignored versions handling

[AMDmi3]

I'm interested in people opinion on handling of ignored versions in case when they are in between devel and stable branches. Here's an example: 2.2alpha1 (latest devel), 2.1.20170101 (ignored), 2.1 (newest stable). The question is what class should be assigned to 2.1.20170101. Possible options are:

• Legacy. Like it's a continuation of devel branch
• Ignored. "Out of sight, out of mind" unambigous solution.
• Outdated. The idea here is that random snapshot is obsoleted by official newer alpha. I like that it does encourage garbage collecting snapshots, but I don't like that it uses some custom non-transparent logic, and it also misbehaves for case of far apart branches, e.g. 3.0beta2 should not really obsolete 2.0.20170101. To fix it, additional version comparison machinery is needed.

[AMDmi3]

On the second thought,

• 2.2beta1 - 2.1.20170101 - 2.1
• 3.0beta1 - 2.1.20170101 - 2.1

are essentially different cases for both legacy and outdated cases. In the second case there's no way we can reliably assume the snapshot is a continuation of devel branch. Seems like we should go with ignored class. Opinions are still grestly appreciated.

[jrmarino]

i vote for "ignored" (in the case there are actual releases).
In the case there are no releases being made, and the dates represent post-last-release commits, then it could be considered "dev".

added: After re-reading your scenario again, I'd still vote to keep it as "ignored". If the author sanctioned the version in question, he/she would have released a patch or a new version. If the version can't be traced back to an actual release from the author, "ignore" it.

[AMDmi3]

Ok, we're going "ignored" way. Actually, I still don't have a real life example for this (e.g. ignored versions between non-ignored ones).