On Wed, 2020-04-22 at 09:48 +0100, Daniel P. Berrangé wrote: > On Wed, Apr 22, 2020 at 01:51:10PM +0800, Shi Lei wrote: > > As you suggested, parsing C-structs plus some magic comments may be a good starting point. > > But parsing C is much more difficult than parsing xml or json. I think this job should base on > > some present tools or other basis. We can first refer to Clang or other light compiler front-end. > > Maybe we can utilize some output of the middle stage of these compilers. > > FWIW, we already have a tool in libvirt that parsers C header files. The > scripts/apibuild.py file extracts info about our enums, structs, APIs and > uses it to build the public API documentation. > > I wonder if we can refactor that tool to extract the code for parsing into > a module, so that we can more reasily re-use it for both the API docs and > for a new XML generator Or we could replace that C parsing code with something based on libclang. Either way, if this ever becomes usable I think it should not live in the libvirt repository but be a standalone tool instead, as I can see many projects potentially benefiting from it. Which begs the question: are we absolutely certain something like this doesn't exist already? We should make sure that's really the case before we invest time on it... -- Andrea Bolognani / Red Hat / Virtualization
