improve SCM detection in Maven mode

[monperrus]

in Maven, the source repo can be filled in:

• url
• scm.url
• scm.connection
• scm.developerConnection

Source: Section 3.2 of AROMA: Automatic Reproduction of Maven Artifacts

[randomicecube]

I've implemented this and looked at what URLs were being acquired now -- from my understanding, the core matter here is that, as per #8 , we need to add support for non-GitHub repositories; most of the repositories do seem to at least point to an existing repo, but they aren't GitHub ones (e.g., gitbox, svn).

I'm assuming that for this, we should compile a list of known/relevant source code repository websites

[algomaster99]

Plus some URLs could be a link to the "project page" rather than source code repository so you will also need a way to differentiate between them.

[randomicecube]

Plus some URLs could be a link to the "project page" rather than source code repository so you will also need a way to differentiate between them.

I'd argue that that's already the case though, no? Which is fair, of course, just trying to understand if it's a new issue arising

[algomaster99]

I just thought I should mention in case you were not aware. The AROMA paper mentions that they find URLs but sometimes it is just project pages like documentation.