On Thu, Apr 21, 2022 at 04:04:20PM +0200, Peter Krempa wrote: > On Thu, Apr 21, 2022 at 06:59:14 -0700, Andrea Bolognani wrote: > > On Thu, Apr 21, 2022 at 03:42:31PM +0200, Peter Krempa wrote: > > > One more thing to consider is (not visible in this snipped patch as we > > > see only a typedef) is that we already do have the version info for > > > functions in the generated API xml. So ... do we even want to be adding > > > them to the comments? > > > > > > Obvious pro is that it's visible right from the function comment when > > > somebody is looking at the code itself. > > > > > > Obvious con is that there are now multiple places that have this info. > > > > I think it makes sense to document version information for all public > > symbols in the same way. > > > > In the case of functions, the other source for this information (the > > symbols file) can still be parsed and used to double-check that there > > are no inconsistencies. > > Sure and it's actually done in this series, which is great. > > We just need a hack for 4 functions where the code was mistakenly added > in a later version than under which the symbol is exported. (also done > in this series, just needs to be fixed). > > And definitely it's more convenient when looking at the code itself than > having to fire up the browser and look at the generated docs or in the > API xml file. Great, sounds like we're in agreement on the approach then :) -- Andrea Bolognani / Red Hat / Virtualization