Runtime and app question

[Phaedrus Leeds]

My understanding is that Philip is just asking for the docs to be changed,
since changing libflatpak to enforce this would be difficult. So I opened a
PR for the docs change: https://github.com/flatpak/flatpak/pull/4105

On Tue, Jan 26, 2021 at 10:45 AM Matthias Klumpp <matthias at tenstral.net> wrote:
>
> Am Di., 26. Jan. 2021 um 18:57 Uhr schrieb Philip Withnall
> <philip at tecnocode.co.uk>:
> > [...]
> >
> > I’d be satisfied with the situation if the libflatpak documentation
> > explicitly said that the ‘kind’ of a ref isn’t part of the identifier,
> > i.e. that refs are a 3-tuple rather than a 4-tuple — see
> > https://docs.flatpak.org/en/latest/libflatpak-api-reference.html#FlatpakRef.description
>
> That would break the libflatpak API though, wouldn't it? As
> `flatpak_ref_parse` would not be able to construct a valid FlatpakRef
> from a 3-part string (and possibly other changes required).
> You could of course at a paragraph stating that the ID part must not
> collide, so there may never be Refs with the same ID but different
> kinds. To actually enforce that though, I guess some deeper changes
> would be needed...
>
> > Other places in the docs may also need to be amended. Some parts of the
> > documentation already omit mention of ‘kinds’ when talking about refs:
> > https://docs.flatpak.org/en/latest/using-flatpak.html#identifiers
>
> Ah, so that's where my assumption that the ID was unique even in
> Flatpak came from! Makes a lot of sense to harmonize that IMHO, but I
> don't know if that could be done as a non-breaking change (for
> Flatpak's API that is, bundles probably don't care at all).
>
> Cheers,
>     Matthias
>
> --
> I welcome VSRE emails. See http://vsre.info/
> _______________________________________________
> Flatpak mailing list
> Flatpak at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/flatpak