This should fix #386
It's maybe a bit harsh to beg the user to upgrade on every opam invocation, but I guess we could improve that with 1.0.0. This should fix #382
…age is the one with max-version. I guess it would be better to give the option to the user to choose which is the corresponding version but no hurry.
… one line).
…eventual issues. This could be extended later by adding more consistency checks.
…this is much more robust (and it not much slower anyway).
…ch is typically the case when we use dev packages), remember which packages successfuly build and install to not have to handle them again on the next upgdrade. This was working well when we were upgrading because of new packages, so now the behavior is much more consistent.
…he root of OPAM (ie. everything goes into ~/.opam/log or ~/.opam/<switch>/build/<package>/)
…ailable. If you want to tweak the stategy used, change OPAMCRITERIA (for instance OPAMCRITERIA=-new,-remove), see /usr/bin/aspcud for available options (the package is available under debian in apt-cudf.
… the solver at the first line of the generated CUDF file (if using --cudf option)
…pparently it can fail on some system and tar simply accepts an output file as argument. This should fix #378