On 7 August 2017 at 09:00, Antonio Ospite wrote: | On Mon, 7 Aug 2017 13:37:39 +1200 Ben Caradoc-Davies <b...@transient.nz> | wrote: | > Is there something wrong with libgsl23 2.4+dfsg-4? nco which depends on | > libgsl23 and {inkscape, qgis} which depend on libgsl2 are not | > coinstallable. Should libgsl23 have libgsl2 in its "Replaces:"? I do not | > understand, but I suspect that something is wrong. | > | | IMHO "Replaces" is not a good solution short-term: | | libgsl2 provides API version 19 | libgsl23 provides API version 23 | | Software depending on libgsl2 might not be ready for API version 23. | | The problem is that libgsl23 depends on libgslcblas0 but the | latter rightfully conflicts with libgsl2 because both packages ship | libgslcblas.so.0
Right. Bummer. Long-term it is of course correct to split libgslcblas.so.0 out as I have done now. | This makes libgsl2 and libgsl23 not co-installable, which makes | the package name change a little less useful. | | A possible solution could be to release a last libgsl2 version (maybe | 2.4 +dfsg-2.1 ?) which does not ship libgslcblas.so.0 but instead | depends on the libgslcblas0 package, also restricting the conflict in | libgslcblas0 to libgsl2 versions *preceding* that version. We can't do it for gsl 2.4 because that already had the missing symbols issue. We can't do a rebuild of the _previous_ GSL either (unless we do ugly fudging with epochs and whatnot). | Long-term there could be a transition to libgsl23 for all the packages | which depend on libgsl2, but I don't know how that is triggered. We may need one. But I am not familiar with the process either. Dirk | Just my 2c. | | Ciao ciao, | Antonio | | -- | Antonio Ospite | https://ao2.it | https://twitter.com/ao2it | | A: Because it messes up the order in which people normally read text. | See http://en.wikipedia.org/wiki/Posting_style | Q: Why is top-posting such a bad thing? -- http://dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org