bug#20637: incompatible, undocumented change to vc-working-revision

[Dmitry Gutov <dgutov@yandex.ru>]

On 04/14/2016 09:31 PM, Michael Albinus wrote:

> Yes. I hope we could use more file properties caches. To be investigated.

OK, that could be a fine solution to the problem of vc-registered's 
slowness, but it adds complexity. So I'm still in favor of equating 
`unregistered' with nil, .

> And as said already several times, if
> we would document vc-* functions in the manual, it would allow us to
> have a more global view on proposed changes.

I disagree. The manual is the documentation for the users, to explain in 
depth, give examples, et cetera. The docstrings and VC's internal 
documentation have to stand on their own. It would be silly if the 
difference between `vc-backend' and `vc-responsible-backend' were to 
only be explained in the manual, but not in the docstrings.

That would also be unfair to people such as myself who prefer to consult 
the latter.

So, do you need anything from me in this area? E.g., feel free to give a 
list of docstrings that seem insufficient to you, together with what you 
feel they are missing.

> I trust you that you have
> all involved interfaces in your mind. I haven't, and I would like to see
> how an interface change compares to the other interfaces.

I don't really know everything about VC, I just have some recollections 
about dealing with it, as well as experience writing a third-party 
package depending on VC's API.

To get an opinion about the current bug report, I still had to dig into 
the code and investigate, look at the commit history, search for call 
occurrences, etc.

> But you have spoken about
> design decisions in the past (for example whether unregistered files
> could be an argument), which I believe is not documented.

BTW, we've mentioned it before when fixing my old bug report about VC 
using too many process calls 
(http://debbugs.gnu.org/cgi/bugreport.cgi?bug=11757#77).

It may not have even been a deliberate design decision, but it's the way 
`vc-state' is used. Which, in turn, allows backend implementations to be 
sloppy in the cases that are (almost?) never exercised.

> And at least for me the "global view" about vc-* functions is missing,
> and how they are related.

I usually tease that kind of information out by reading the source code. 
Is there anything in particular I could help add to your understanding 
of the "global view"?

> Yep. Pls test my patch, and confirm whether it is sufficient. Same for
> Glenn, if possible. I would like to close this bug then, removing a
> release blocker for Emacs 25.1.

It must fix this bug, since it reverts to the old code, and testing 
Glenn's example from the description confirms as much.

So I think it can be closed, and the discussion should move to emacs-devel.