ORT include concept for monorepo

[nnobelis]

We have a gigantic monorepo containing several modules.

Our users want to scan only one module corresponding to the software they are delivering.

ORT excludes are not scaling too well for such use cases as you have to write a lot of them. Additionally, when a new directory is added to the monorepo, it is taken by default.

Has the ORT team ever considered the concept of ORT includes (allow list) i.e. everything is ignored by default and you have to specify what do you want to include in your run, for instance with some ORT includes ?

[sschuberth]

Have you alternatively considered to limited the initial cloning of the project to scan to certain sub-directories via partial clones / sparse checkouts?

[sschuberth]

Paths could be globs, and also at least Git underneath supports multiple paths. You'd just need to ensure to supported that in the CLI as well.

[nnobelis]

To be sure I understand you right, you mean put a glob in the vcs-path and use the downloader ?

[sschuberth]

Yes.

[nnobelis]

I will have a look but I am not convinced. I have the guts feeling that a lot of different directories must be selected, so putting everything in a glob would not scale well.

[sschuberth]

You could also make --vcs-path take a comma-separated list of globs, if that helps.