Open AltGr opened 9 years ago
yes, the conflicts regarding cstructs are mostly related to self conflicts. If I understand correctly, the idea in #1703 is that since self-conflicts are "implicit", they should not be shown to the user. But if a package fails to install because of a self conflict, what kind of explanation should we give to the final user ?
After some more thought, I get it: doesn't sound easy to explain better indeed... here the real issue must be something like:
the full explanation would be to lift the conflict, but we would have "b conflicts with d" implicitely ;
or maybe more clearly, b conflicts with c <= v
(or equivalently d conflicts with c >= v
) ...
Not sure introducing implicit conflicts is a good idea, but I can't find a better explanation here.
I thought some more about these conflicts:
b
depends on (c > v)
" actually implies "b
conflicts with (c <= v)
"It would actually seem fair and consistent that this is treated in the same way, as a conflict, in dose and the reports. Would that make these results clearer, though ?
At the end of http://ows.irill.org/latest/today/summary.html#conflicts , conflicts between different versions of cstructs are shown, which is weird. Possibly the same issue as https://github.com/ocaml/opam/issues/1703, which would take its root in dose ?