[PATCH wayland v2 0/1] protocol: Add summaries to event parameters

[Bryce Harrington]

On Mon, Apr 04, 2016 at 04:56:04PM -0500, Yong Bakos wrote:
> > On Sat, Apr 2, 2016 at 2:48 PM, Yong Bakos <junk at humanoriented.com> wrote:
> > From: Yong Bakos <ybakos at humanoriented.com>
> > 
> > This v2 patch incorporates some recent feedback per v1. Two things to note:
> > 
> > 1) The 'name' parameter in wl_registry events has been changed to "numeric name"
> >    to indicate that, "yes, this is numeric but it serves as a name."
> >    This is a "best option" measure to put /something/ here to squelch the
> >    warning, which is the main goal of this patch. The semantics of the name
> >    'name' is another concern that I feel does need addressed, but not here.
> > 
> > On Apr 4, 2016, at 2:23 PM, Bill Spitzak <spitzak at gmail.com> wrote:
> > How about "server's id"?
> > 
> > I don't think the word "name" should be in the description at all, as it strongly implies that the data is a string.
> 
> 
> I agree that this needs addressed, "but not here." For example, pq has asserted
> that the name argument is not an id. While I don't entirely agree, I'm a noob
> and I feel more thought needs to go into how this argument is used, what values
> are assigned to it, how the name property of the corresponding struct is used
> elsewhere, what to rename it to, and whether or not it's worth the
> repercussions. The description will be the easy part. I am only now digging
> into this, so if you already know the answers, please educate me!

Yes, that's a sensible approach.

> I'd like to get this patch merged to suppress the `make check` warnings, and
> then submit an intelligent patch that addresses this 'name' parameter
> everywhere it occurs.

I'm unclear if renaming a parameter might constitute an interface
change; if so, then it likely isn't going to be worth the hassle to
change this.  The parameter description such as you're proposing can be
adequate to remove the ambiguities that might exist.

Bryce