On 2018-07-31 23:40:40, Andreas Beckmann wrote: > > Replaces: libpqxx6 (>= 6.2) > > Breaks: libpqxx6 (>= 6.2) > > That is what I meant. > (Usually you would use (<< 6.2.4-3) if you moved some files, but here > you uploaded two buggy package versions (-1 (wrong library name) and -2 > (wrong package name)) while the older versions are OK and > co-installable; and the old package name "disappears" in -3.) > > > To be honest I'm not sure if this is what we need as libpqxx-6.2.so has been > > introduced in 6.2.4-2 and 6.2.4-3 is broken for upgrade. If I'll upload > > 6.2.4-4 > > then I'd expect to replace all earlier versions of 6.2, > > What do you plan to change for -4? > The renaming of the package to libpqxx-6.2 sounds like the correct > solution to me.
This has been done in libpqxx-6.2.4-3 and changes planned are to add B+R only,
plus do the follow through on the transition and binNMUs.
> You cannot have a libpqxx6 package not shipping
> libpqxx-6.1.so . This just needs to be handled as a proper library
> transition (you already filed the corresponding bug).
>
> > so I'd expect B+R as
> > below:
>
> In which package?
It was meant for libpqxx-6.2 but now I see where and why it's going to be
problematic.
> > Replaces: libpqxx-6.2 (<< ${binary:Version})
> > Breaks: libpqxx-6.2 (<< ${binary:Version})
>
> This sounds seriously wrong. (Unless you have a very special package, or
> a very big strange mess to be cleaned up.)
> Why would you want to B+R the newly introduced package?
> Why would you want to bump the B+R every time you upload a new version?
> You need Depends: foo (= ${binary:Version}) to keep something in sync
> but for B+R there is usually a first "good" version that no longer needs
> to be replaced, so no substvar needed.
> (The exception would be a package frequently getting new upstream
> releases in stable with diverging packaging in stable and sid, where
> every upload to stable would invalidate the first "good" version in sid.
> cf. chromium)
After looking into this situation after not so bad night sleep I have to say
that I think you're correct and your explanations are also sound. Thank you for
your help.
I'm going to use B+R accordingly to what you're suggesting.
--
|_|0|_| |
|_|_|0| "Panta rei" |
|0|0|0| -------- kuLa -------- |
gpg --keyserver pgp.mit.edu --recv-keys 0x686930DD58C338B3
3DF1 A4DF C732 4688 38BC F121 6869 30DD 58C3 38B3
signature.asc
Description: PGP signature
