Please do not focus on
rclone: that was an example. The item I wanted to report is not about
rclone that incidentally takes a nice
--version option, but with the fact that
pamac search may mislead its user to think that one version of a package is installed from the distro repo, when in fact a different version is installed locally.
Typically when people get confused, they won’t even realize that they are confused and ask. They might just take wrong decisions (e.g. reporting a bug upstream) based on their confusion.
The problem is not
No need to be suspicious. That was not an anonymous source. The package came from a local build (practiced with Manjaro’s
buildpkg tool) based on the official updated PKGBUILD used for unstable.
Should be exactly the other way round. If you take a single package from Testing or Unstable, things may easily fail because that package might have been built against different dependencies (also from testing/unstable) than the ones you currently have. Conversely, by taking the new PKGBUILD and building locally by
buildpkg you assure that the package is built against your current packages from stable, namely you do what is commonly called a backport much less likely to fail.
Please, no, do not insist on the fact that you should move a whole system to unstable just because you need to quickly fix a single package via a backport. It is not relevant to what is being reported either.
I understand that backports are way less important on a rolling distro than in a non-rolling one, but when there is a need for it, the solution should be to temporarily build and install the backport locally (until the rolling distro has time to catch up), not to make the whole of your system more likely to break by switching to a more unstable rolling model (particularly if your system is multi-user).
To summarize: it was not my intention to create a long thread about
rclone or the opportunity to switch a whole system to unstable.
What I merely wanted was to report a glitch in
pamac (just assuming it might have been of some interest to the
pamac developers, which I now tend to assume is not the case).
pamac appears inconsistent about package versions when you install packages built locally:
- if you install locally package A at version Rxxx for which there is no counterpart in the distro repos, then
pamac search A will correctly report that A is available at version Rxxx and installed;
- conversely, if you install locally package B at version Rxxx for which there is version Ryyy in the distro repos, then
pamac search B will report that B is available at version Ryyy and installed, when in fact it is installed at version Rxxx.