Source code availability per every landing, using ZFS snapshots of source-archives
Alternative location (mirror) for downloading upstream source code archived tarball for oi-userland is:
http://dlc.openindiana.org/oi-userland/source-archives/ ( which can be used automatically if you set the EXTERNAL_ARCHIVE_MIRROR variable.)
Proposes for Openindiana to have source code availability , that are accesible per-every landing and refresh in public repositories.
Zfs dataset(s) holding oi-userland/source-archives/ can be zfs snapshot -ed on every osnet-incorporation and userland-incorporation update,
so that the source code for every landing in /hipster repository from Jenkins, also have same-named snapshot, that is also available from the outside world.
If there is email@example.com,5.11-2017.0.0.8925:20170519T221014Z , to have all upstream source code available for that state of packages, there only needs to be the zfs snapshot:
that is made available under:
Additional repositories to snapshot on build:
That could also work for osnet-incorporation , where osnet-incorporation specific sources and other packages that build with osnet-incorporation could be on oi-osnet/source-archives/
and for hipster-encumber repository in oi-encumbered/source-archives/
Zfs snapshots of previously used source states for building packages during binary repository update process between snapshots,
could be destroyed, after in-between binary builds have been made unavailable/deleted.
Also some of the snapshots can be later cloned, to form source base for future supported branch, also without using additional space.
Deletion of sources zfs snapshots , between 2 OI snapshot source states, go to be careful, not to destroy also snapshots for sources for OI snapshots, because OI snapshot ISOs and .usb images are going to be available for a long time, so source code should be also available.
All this is done because binary distribution gives burden to those distributing built binaries, to also distribute the source code at all times and at the same time with the binaries distribution, to satisfy colypeft licenses used by most of the packages.
This snapshot-and-make-available step, as the part of Jenkins automated publishing, is painless, can be set to be automatic, be low resources and is relatively easy to implement to solve the burning problem with mandatory source availability of binary-distributed copyleft source components, per their respective licenses.
All source patches made to existing sources inside OI project and available on Github, can also land in respective source archive datasets.
But that would also require changing of a build process by putting archives first in folder and then building the component before publishing, but that could require build process changes, and is not the part of this Feature request.
Also available in: Atom