[Tagging] Mapping nonexistent paths

[ael]

On Mon, Mar 22, 2021 at 06:09:26PM +0300, Bert -Araali- Van Opstal wrote:
> 
> On 22/03/2021 15:22, ael via Tagging wrote:
> > On Sun, Mar 21, 2021 at 04:06:37PM -0600, brad wrote:
> > > A nonexistant path shouldn't be mapped.
> > Justification? That seems just dogmatic. A virtual path (or some
> > equivalent) solves a longstanding problem with routers failing to cross
> > accessible open ground.
> > 
> > If someone can walk across a space in a safe and legal way, a "path"
> > comes into existence, at least in the abstract.
> > 
> > I do think that "virtual" rather than "visibility=no" conveys that
> > situation more clearly.
> > 
> > ael
> So are you actually saying we should map "virtual", "non-visible", in
> essence non-verifiable things ?

No. This is verifiable. I can survey to determine that it is possible to
walk across an area. It is a possible path for me.

> Map for the router?  A router is a piece of software, some less some more
> sophisticated, the more sophisticated ones could suggest where to pass.

I am generally against "mapping for a renderer". But that means
adjusting tagging to compensate for a perceived or real deficiency of a
particular renderer, rather than fixing the renderer. We are (almost)
always mapping for renderers and routers and possibly for other uses in
the general sense: otherwise there is no point in mapping at all. So
yes, absolutely tag for the router(s) where that is sensible, and, as
here, in a verifiable way. There is no need to overcomplicate routers
with inference engines to deduce where open space can be crossed which
will almost certainly get it wrong in many cases when we can have a
simple tag to assist.

ael