On Wed Mar 19, 2025 at 08:10:19PM +0000, Stuart Henderson wrote: > On 2025/03/19 16:55, Theo Buehler wrote: > > It mostly looks good to me ports-wise, although I'm not at all convinced > > by the devel/bear use case. It feels very expensive for what it provides. > > I guess I'm just not the target audience. That's fine. > > > > Do we really want or need to ship these, especially the systemd one? > > > > lib/cmake/grpc/modules/ > > lib/cmake/grpc/modules/Findc-ares.cmake > > lib/cmake/grpc/modules/Findre2.cmake > > lib/cmake/grpc/modules/Findsystemd.cmake > > I wonder if they might reference each other, maybe I'll take a look > if I remember to do so when it finished building ;) > > > I am really reluctant okaying this just because it is yet another Google > > C++ monster that will significantly add to bulk build times. > > (especially on archs other than amd64/aarch64) > > > I can live > > very well without this and its only prospective consumer. On the other > > hand, nobody's really objected so far... > > Same for me. The other thing with Google C++ monster libraries as far > as I've seen is that they often don't seem to concern themselves too > much with cross-version compatibility, I think they may assume that > the typical downstream user will vendor them. > > > What do others think? Yay or nay? > >
Another option would be to hack around the ExternalProject_Add(grpc_dependency.. part in bear(1): https://github.com/rizsotto/Bear/blob/master/third_party/grpc/CMakeLists.txt#L27
