Syntactic problems with specifying a rare map layout (not a code bug report!)

[gwselke]

For a map that I am currently working on, I find a curious state of affairs in The Real World which I struggle to accommodate in the Map::Tube map syntax. (This is for XML, but I think the same applies to JSON.)

Here is a stripped down version of the situation:

Line X (id x): A <-> B <-> C
Line Y (id y): A -> B -> C -> D -> A (unidirectional; no direct link (neighbour) from B to A)

The best I can come up for the map data is then:

<station id="A" name="A street" line="x,y" link="B"/>
<station id="B" name="B street" line="x,y" link="A,C"/>
<station id="C" name="C street" line="x,y" link="B,D"/>
<station id="D" name="D street" line="y"   link="A"/>

But here the information is lost that, while there is a direct link from B to A using Line X, there is no (direct) link from B to A using Line Y. Consequently, get_shortest_route( 'B street', 'A street' )->preferred will yield
B street (LineX, LineY), A street (LineX, LineY)
instead of
B street (LineX), A street (LineX)

I do not think this is a bug in Map::Tube::get_shortest_path( ) or in preferred( ), since the result is compatible with what the map says. The issue is rather that the map does not say what I need it to say. And I have found no way of making it say what I want it to say.

This is certainly a rare edge case, and the consequences are very mild (the result being just slightly off). I am not sure whether such a case can be modelled given the current XML (or JSON) syntax. But changing the syntax would almost certainly break many existing applications, and that should certainly be avoided. Also, the effort may be too big for little gain. So this should not be regarded as a bug report, but as food for thought -- maybe there is an easy way to express this, after all.

[manwar]

@gwselke just wanted to confirm with you, if this is acceptable route using the above map?

   B (X), A (X)
   B (X, Y), C (X, Y)
   C (X), B (X)
   A (X, Y), B (X, Y), C (X, Y)
   A (X, Y), B (X, Y), C (Y), D (Y)
   A (X, Y), B (X, Y)

[gwselke]

B (X), A (X) - correct
B (X, Y), C (X, Y) - correct
C (X), B (X) - correct
A (X, Y), B (X, Y), C (X, Y) - correct
A (X, Y), B (X, Y), C (Y), D (Y) - Almost. A (X, Y), B (X, Y), C(X, Y), D(Y) for the result of get_shortest_route( 'A', 'D' ). For preferred( ), the result should arguably be A (Y), B (Y), C(Y), D(Y)
A (X, Y), B (X, Y) - correct

[manwar]

Thanks @gwselke
I will get that sorted soon.
In the meantime, if you have any other test routes I should be looking at, then please do share.

[gwselke]

Please, have a look at #24. I just documented a similar path problem there, using the test data that I have provided for that issue.
(The main topic of issue #24, however, is not route-finding but the display of lines as a diagram.)

[manwar]

@gwselke I got all the above routes working as expected by you.
It's a real hack and needs proper review.
The best part is that it's not a blocker currently.
We have all the time in our hand to keep polishing the logic until it is production ready.