Quoting Lev Lamberov (2020-04-23 14:31:32) > Чт 23 апр 2020 @ 14:15 Jonas Smedegaard <jo...@jones.dk>: > > >> Ohhh, I see. Let me upload (source-only) 8.1.29 to unstable. Will > >> be OK for you? > > > > If I understand the situation correctly, then this is a bug in how > > eye is packaged: A prolog image is dumped during build and shipped > > as the executable for eye. That image makes starting eye faster, > > but requires same version of prolog. > > > > My current thinking is that no change is needed from prolog: > > Instead, eye must be tightened to depend on same version of prolog > > as was used to build the image (and then each new release of prolog > > needs a binNMU of eye). > > > > Does that sound correct to you? > > Looks like you're correct.
I think I have a fix prepared now. If you happen to know that the image format it stable across some releases of prolog then it would be more legant if the prolog package were to declare its ABI - e.g. by a "Provides: swi-prolog-abi-dead1234" in control. But I guess there is no such thing. > Another thing that disturb me is that eye downloads test data from > github repository while testing. So, change in tests upstream may > break stuff in Debian even without any change in the Debian archive. > But maybe I'm wrong about volatility of upstream test data. What do > you think? I agree it is worrisome. That test is annotated as "flaky" for that reason - and from next release will also be marked as "needs-internet". Would be better to put together some local tests, I just haven't taken the time to do that yet. Help is much appreciated :-) - 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
