The right way to go.
- you propose with your mail to update and upload the v2
src:ruby-aws-sdk package to unstable (an earlier broken(?) version is
in experimental)
As tests say, its broken, yes.
- we don't discuss yet the upgrade to v3, but it will be needed at some
point because some rails apps need them (loomio).
I dont have any problem with loomio... I have a problem uploading things that brokes users functionality.
If other parties involved in packages using ruby-aws-sdk are ok, I would
be happy to help you get this v2 to unstable.
The only reverse dependcy of aws is gitlab, and the current version is totally broken.
Is strange, that a reverse depency of gitlab, that is taking too much love on his dependencies, has one totally broken...
It would be faster to jump directly to v3, but there are some issues:
- the multibinary layout can help you create a source package from the
github repo
- but it would result in a huuuuge quantity of binary packages. It is a
lot of work for FTP masters to review them (once) and additional load
on the archive to add so many packages
As I say, packaging from gems should not the way to go. Is like packaging C libraries from distribution tars (not source ones):
distribution tars == binary ones
I dont see any complains on the review by the ftp masters... Nor on the load on the archive...
We discussed this issue a little bit during the sprint, and I kind of
remember that the proposition we had was to have this multibinary source
with only the needed services provided as binary packages. Was it the
statement we reached? Dear participants of the sprint, don't hesitate to
say I am wrong...
Cheers,
Cédric