| Summary: | allow overlay-wide RESTRICT=mirror in metadata/layout.conf | ||
|---|---|---|---|
| Product: | Portage Development | Reporter: | Sergei Trofimovich (RETIRED) <slyfox> |
| Component: | Enhancement/Feature Requests | Assignee: | Portage team <dev-portage> |
| Status: | RESOLVED INVALID | ||
| Severity: | normal | CC: | cynede, esigra, jstein, kensington, kingjon3377 |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | |||
| Bug Blocks: | 240187 | ||
(In reply to Sergei Trofimovich from comment #0) > metadata/layout.conf: > restrict = mirror <anything else> That sounds reasonable. We could also allow negative incrementals in ebuilds, so that things like RESTRICT="-mirror" are possible. How RESTRICT="-mirror" should work? restrict RESTRICT=mirror in overlay ebuilds? The restrict setting from metadata/layout.conf would provide default RESTRICT settings for all the ebuilds in the overlay, and the ebuilds could override the defaults. They would set RESTRICT="-mirror" to remove mirror from the current ebuild's RESTRICT, so it would not inherit mirror from the defaults. I think it is important to keep this information in the ebuild itself. Else it could get lost too easy by copying ebuilds and false RESTRICT information is a) a legal problem and b) very difficult to detect for us. The original use case had nothing to do with license restrictions. Nothing prevents you from not putting RESTRICT=mirror to the ebuild in ::gentoo and be in trouble. Or having a bad license. As for original request ::haskell ended up actually having a mirror for overlay as well. |
haskell overlay has around 1000 packages of them only 200 are in main tree. Thus most of fetches from overlay first hit main gentoo tree and then fallback to SRC_URI. What do you think of allowing to mess with it right in layout.conf? metadata/layout.conf: restrict = mirror <anything else> [ Don't know if it also needs to be addressed in PMS ] Thanks!