Quoting Lev Lamberov (2020-06-07 08:31:12) > Hi Jonas, > > Сб 06 июн 2020 @ 23:02 Jonas Smedegaard <[email protected]>: > > > Quoting Lev Lamberov (2020-06-03 12:20:14) > >> Пн 25 мая 2020 @ 12:32 Jonas Smedegaard <[email protected]>: > >> > >> > Quoting Lev Lamberov (2020-05-25 12:19:07) > >> >> I've just got some news from Jan Wielemaker. The next stable > >> >> release of swi-prolog, branch 8.2, is almost ready. How are your > >> >> experiments with depending on swi-prolog virtual packages going? I > >> >> would like to package the current latest version and upload it into > >> >> unstable for testing it with more installations, just to make sure > >> >> there are no serious bugs and blockers for the release of 8.2 > >> >> branch. > >> > > >> > Sorry, I have not played with it at all, yet. > >> > > >> > Hope to find/make time for it later today. Will let you know when I > >> > do! > >> > >> A couple days ago I uploaded new stable release of swi-prolog, 8.2.0, > >> into unstable. This branch of swi-prolog will stay for a long period > >> of time. Quite possible that it will be in Debian stable, since I > >> don't think we'll see another stable release of swi-prolog before the > >> coming freeze. > >> > >> Could you please upload eye rebuilt against swi-prolog 8.2.0? > >> Currently swi-prolog are going to be removed from testing on June 22 > >> because of RC bug in 8.1.29, and there are some other packages > >> depending on swi-prolog which are going to be removed too. > >> > >> Also, it would be nice if you can switch dependency on swi-prolog to > >> one of its virtual packages. There are > >> swi-prolog-abi-2-67-4369f15b-de23899e and swi-prolog-abi-foreign-2 > >> provided by swi-prolog-core. You may find more information about ABI > >> versions in its NEWS.Debian and mentioned there upstream > >> documentation. > > > > Eye updated now, using the new virtual ABI package, but unfortunately > > couldn't move to lighter dependency than the -nox package as that has a > > needed pcre module. > > > > I hope others will benefit from the package split. Speaking of which, I > > suggest to have the core packages only suggest (not recommend) the -doc > > package. > > Thanks! > > Guess we still need to ask for force migration, since as I can see eye > is marked as migrating _after_ swi-prolog. > > Also, it would be nice to figure out which part of swi-prolog ABI > version matters for eye. As Jan wrote there are > > > 2-67-792e14f8-de23899e > > > > - Backward compatibility version of the foreign interface is 2 > > - Saved state file format can be loaded when not older than 67 > > - 792e14f8 is a fingerprint for the C-defined foreign predicate > > signatures. > > - de23899e is a fingerprint of the VM instructions and their > > signatures. > > IIRC this version are for 8.1.30, and for 8.2.0 (currently in Debian) > the version is 2-67-4369f15b-de23899e, that is foreign predicate > signatures were changed. > > IIRC eye contains some saved state files to make it run faster. So, does > just saved state format matter for eye? If so, I'll make swi-prolog-core > provide something like swi-prolog-abi-states-${...}. Cause the latter > hashes are rather volatile, if I'm correct. > > Jos, Jan, could you give a hint on what matters for eye and which kind > of Provides is better to have for it? The goal is to make swi-prolog > updates possible without the need to rebuild eye each time, but rebuild > still should be required in case there are some incompatible changes in > swi-prolog (which should be reflected in its ABI).
I think one step towards generally more flexible packaging would be to not offer 2 virtual ABI packages (foreing and all) but 5 (each component and all). It would also help if you documented in README.Debian when how to make use of those virtual packages (i.e. a more long-term oriented place with all details, that you can point to from NEWS instead of pointing to our bugtracker). Thanks, - Jonas -- * Jonas Smedegaard - idealist & Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ [x] quote me freely [ ] ask before reusing [ ] keep private
signature.asc
Description: signature
