https://bugzilla.redhat.com/show_bug.cgi?id=1495985 --- Comment #8 from Sergey Avseyev <sergey.avseyev@xxxxxxxxx> --- Thank you, Igor. Your notes are very useful. I applied almost all of them. > 2. The conditional doesn't make sense because F21 is EOL for long time, so I would recommend you to remove this conditional at all I included this to reuse the same spec in future for EPEL, but I think it would be better to keep it clean for now. >> -DLCB_BUILD_LIBUV=OFF > Why not to build libuv part as well? We have it in Fedora.. We didn't do it previously in our builds, because in the past libuv was not common to be shipped as library, but I think it make sense. I will update spec to include libuv IO plugin too. > And you probably want to add soname for libev / libevent libs, not sure how it is supposed to work though (see https://fedoraproject.org/wiki/Packaging:Guidelines#Downstream_.so_name_versioning). I saw that issue in rpmlint, but currently we don't use versioning for the shared objects of the plugins, and check their versions in runtime. Ideally these plugins should not be placed in the libdir at all, and should be loaded from other place like %{_libdir}/%{name}/libevent.so. I plan to improve that in next major version in the upstream. If it is critical and blocks accepting package into Fedora, I think I can create patch to workaround it. Right now in upstream I'm doing something simple like: dlopen("libcouchbase_libevent.so", RTLD_NOW | RTLD_LOCAL); Updated artifacts: Spec URL: https://github.com/avsej/libcouchbase.spec/raw/a8c1b09ba9afeaea67c657ce85fc1cd06f5170b4/libcouchbase.spec SRPM URL: https://github.com/avsej/libcouchbase.spec/raw/a8c1b09ba9afeaea67c657ce85fc1cd06f5170b4/libcouchbase-2.8.1-1.fc28.src.rpm Koji build: https://koji.fedoraproject.org/koji/taskinfo?taskID=22089176 In the next step I'm going to add package libuv IO plugin, and test if the plugins could be easily loaded from "%{_libdir}/%{name}/" -- You are receiving this mail because: You are on the CC list for the bug. You are always notified about changes to this product and component _______________________________________________ package-review mailing list -- package-review@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to package-review-leave@xxxxxxxxxxxxxxxxxxxxxxx
