Measuring Wine's API completeness

[Alexandre Julliard]

Michael Jung <mjung at iss.tu-darmstadt.de> writes:

> In my opinion it's not a good idea to interpret the absence of stub or 
> semi-stub as 'implemented': I guess very few of wine's APIs would be 
> considered implemented, yet most of them don't have a 'stub' or 'semi-stub' 
> annotation. This would lead to a level of false positives, which would render 
> the result almost meaningless.

It's a rough estimate, but it has the advantage of not requiring code
changes. Adding special comments all over the place is going to be a
lot of work, maintaining them properly will be even more work (and
most likely won't happen on most functions), and despite all that
extra work the final stats will be just about as meaningless.

If you want to measure the quality of the implementation a much better
way IMO is to measure regression test coverage. This may also motivate
people to write more tests, which would do a lot more good than
spending time adding magic comments all over the source.

-- 
Alexandre Julliard
julliard at winehq.org