On Fri, Feb 21, 2020 at 06:49:06AM +0000, Gordon Ball wrote: > On Thu, Feb 20, 2020 at 09:31:16PM +0100, László Böszörményi (GCS) wrote: > > On Thu, Feb 20, 2020 at 9:06 PM Luca Boccassi <bl...@debian.org> wrote: > > > On Thu, 2020-02-20 at 20:57 +0100, László Böszörményi wrote: > > > > Luca, what's your opinion on this? I think that needs to be packaged > > > > separately. > > > > > > It's a bit weird, but given it's just a header, IMHO it's fine to keep > > > shipping it in libzmq3-dev. Upstream we generally try our best to keep > > > the hpp header forward and backward compatible, so there should be no > > > surprises. > > > I do the same in the packages I build upstream in OBS. > > OK. I don't see the point then being developed in a separate git tree > > - but let it be. I'm going to sync it with zeromq3 packaging. > > Thanks for responding to this quickly. > > I don't have a strong opinion on whether the headers from the cppzmq > respository should be bundled in libzmq3-dev or split out into a > separate package (although if they're largely released in sync, having a > separate package for two header files seems unnecessary). > > However, the original request in #951135 was for `zmq_addon.hpp` from > `cppzmq`, which has not been included in 4.3.2-2. Was this intentional?
Sorry, ignore me - I didn't see mention of `zmq_addon.hpp` in the changelog and then failed to notice that the file list on sources.d.o was out-of-date and still showing 4.3.2-1. > > (To be clear, this isn't at all urgent - I'm just experimenting with > xeus at the moment. I encountered this issue because I assumed cppzmq > was unpackaged and made a local stub package to install it, only to find > it conflicted with libzmq3-dev. Assuming that the package will continue > to ship the C++ headers, adding the string `cppzmq` to the description > or `Provides: cppzmq-dev` might aid discoverability). > > Gordon > > > > Regards, > > Laszlo/GCS
