Hi Peymaneh, On 8/6/21 11:03 PM, Peymaneh Nejad wrote: > > Hi Nilesh > >>> - I have excluded a couple of modules (like cmd) that were introduced since last version, cause trouble when building and are not needed for purpose of packaging caddy. > >> ^^^^^^^^^^^^^^^^ > >> I understand that packaging caddy is important for you, but I think it is also important to keep in mind that we can not go about breaking the whole world just because of that. > > > ACK. > I am not sure however if there is a way *not* to break the whole world in this situation, maybe I should go into detail: > One module that I have excluded depends on github.com/google/cel-go and would introduce a circular dependency, since cel-go depends on grpc itself. grpc already has one circular dependency. > Exluding cmd/ seemed reasonable to me because grpc ships no binaries and the tests for cmd/ would cause sbuild to freeze, and shipping the source files for a main function that does not even have tests and introduces even more dependencies seemed to me to serve no purpose. Thanks for the details, makes things clearer now. I see. But In such cases, it is better to discuss w/ maintainer and more folks, and do stuff with the best way, IMO The thing is, I guess uploading to experimental would be a good thing, but if it is done in a state that is close to what would go to unstable, that makes situation happy for everyone. The problem is: a) It might need a major overhaul to get to unstable, creating more work (and we are pretty close to a release too) b) If something $new and unexpected is introduced, and it breaks caddy, we will spend time in fixing it later, rather it is easier to do it now Right? >> The problem I see with the current state is that it will create additional work for the maintainer to include those things in while doing transition. grpc is a very important package with a number of reverse > >> dependencies. In principle, uploading to experimental should be harmless, but I think we should consider to ask the maintainer once about it. > > Yes that makes sense. I would try to fix the new version at least as far that it does not cause any regressions, and ask its maintainers what the think of my changes. > > What do you think? Yes, that sounds sensible Nilesh
Attachment:
OpenPGP_signature
Description: OpenPGP digital signature