epb_compatibility overrides module_name_suffix

[tsloughter]

I use epb_compatibility to tell gpb to generate the encode/1 and decode/2 functions but now it is also forcing the suffix _pb even if i set {module_name_suffix, "'}. I don't think this is proper, is it?

[tsloughter]

I'm currently using a work around since I realized that if I simply define {module_name_suffix, ""} before epb_compatibility it will use that value instead of "_pb" :)

[tomas-abrahamsson]

(Yes, was going to suggest that workaround as a short-term solution :) )

Hm. I was assuming the Erlang protobuffs always suffixed with _pb, so it'd be a natural thing for an epb_compatibility option to do, maybe even an expected thing of such an option.

Perhaps a sensible way forward would be to add (yet) an option epb_api or epb_functions for the functions, and let the epb_compatibility imply this and the other options outlined in #63. That way, you could specify epb_api (or epb_functions whichever is chosen) instead of epb_compatibility.

You mentioned earlier that you were a mix of erlang_protobuffs and custom code. Is this lacking of _pb suffix part of the custom code? Or am I wrong in my assumption that _pb is always added? (if yes to the latter, then scrap the option idea above)

[tomas-abrahamsson]

I've got some code locally in which generation of functions encode/1 and decode/2 is now controlled by an epb_functions instead, which you could use. (And the epb_compatibility option expands to epb_functions plus a few more options)

But I'll wait a bit before pushing that, to hear your story, because I'm still confused about erlang_protobuffs and yet no _pb suffix, because maybe it means some other solution is better.

[tsloughter]

Hm, you seem to be right. We weren't actually using epb as a dep but the modules from it for generating the encode/decode functionality and writing it to a file of our own.

So it is only our use case that made this funky :). I'm fine with the workaround hack and we can close this.

[tomas-abrahamsson]

Ok. I'll close this, but I'll probably push the epb_functions option in a while anyway, maybe waiting a bit first to see if any more changes pile up. Your workaround will continue to work the same way even with that, I figure.

(I also learned something about option design: if an option is a meta-option that expands to other options, then it should only expand to other options, and not have any extra meaning of its own; such extra meaning should be broken out to separate options instead)