canqc: Quebec Provincial Highways

[yakra]

No objections on how the QC125 split was handled. It's an imperfect situation begging for an imperfect solution.
That does it for my comments.

[oscar]

QC185/TCH
...
 - is the QC185/A-85 point in the right place? Is this one-point-per-interchange

Assuming you mean A-85_S, it's that we decided in CHM days after the painful Arnprior experience with ON 17/ON 417 to end Canadian freeways at the last interchange, unless there is an intersection point where we could place the end (as we did with the other A-85 segment, where there is a Fin A-85 sign at 3e Rang). There is signage indicating that A-85 restarts before sortie (exit) 47 in St-Louis-du-Ha! Ha! But there is no intersection between there and ChSav. ChSav has signage indicating that QC 185 extends east of there, and A-85 doesn't end there, for now.

That all being said, it looks as if Exit 47 would be better served as a "misbehaving parclo" to the west.

I don't see the advantage, especially since the RIRO ramps from/to SB A-85 are about 0.25 km north of the NB ramps. Using either as the 47 point location would add up to 0.8 km to canqca and subtract it from canqc ... for now, since Transports Quebec has begun construction on filling in the gap between the A-85 segments, which in a few years will fold all of QC 185 into A-85.

QC309: Is ChLaj semi-important or likely to be used by travelers? If not, Chemin du Rubis to the east would do a better job of shaping.

ChLaj is adequate for shaping purposes. It also is, or connects to, a much longer road than Ch. Rubis which dead-ends in less than 2 km.
 

QC342: MteeBStTho -> something else

MteeBaieST?

QC360
 - QC138_A, QC138_B, QC138_C, QC138_D -> QC138_Que, QC138_W, QC138_E, QC138_Che ?

Despite the change having already been made, IMO with these point all connecting to the same chopped route, _A _B _C _D might be the less ambiguous option.

My initial preference was _A, _B,  _C,  _D. But I deferred to si404, and would rather stick with that, even though changing back would not be difficult for either QC 138 or QC 360.

QC138Mon
 - USA/CAN is slightly off (also NY30)

The point is a bit N of the border per OSM, and a bit S per ESRI.
What do NYS shapefiles have to say?
The MilepointRoute2015 and Cities_Towns shapefile sets are at different scales / different levels of precision. Consequently:
• The arc representing NY30 crosses the edge of the Constable polygon here, corresponding to where OSM has the border.
• The northernmost extent of the NY30 arc is here, fairly close to where the current waypoint is, but still a bit south of where ESRI has the border.
In light of all this, I'm fine with going no-build. Or I could change to something closer to OSM if Oscar wants to.

Thanx for checking. I would leave this one alone.

A-20(63) -> 4

Oh dear. And since this qualifies QC138 as a route with its own exit numbers, possibly A-20(64) -> 64(20)?
Does this happen elsewhere in the system?

I'll change 4 back to A-20(63).

As for other examples, I'm scratching my head on that one. The one that comes to mind is QC 116 near Montreal, which has its own exit numbers, including on a segment concurrent with QC 112.

Also, the points for the north end of US 11/south end of QC 223 look off relative to the border as shown in OSM, and to a lesser extent in ESRI (including its satellite imagery) and HERE. Do shapefile data support leaving the border point as is?

Similar situation to NY30. I get coords here and here.

The second set of coords are pretty close to the existing USA/CAN coordinates. The first are close to the border as shown in OSM. I would say the "shapefile data support leaving the border point as is". No action needed.

BTW, there are possible isolated segments of QC 138 in La Romaine (at least from the harbor to the airport), Chevery, and Pakuashipi, some of which reportedly are signed and one of which (Chevery) is in OSM. Everything I've seen in MTQ online data shows route numbers for those segments other than 00138, so I'm inclined to treat them as "future 138" at most. @yakra, do the most recent shapefiles say anything more?

Nothing more recent. Still working with data from 2016.
GeoBase seems to have slowed down big time on, if not forgotten entirely, updating the NRN shapefiles. NS & PE were updated in 2018; everything else...
(Maybe I gotta dig up a new link? I've tried occasionally, with little success...)

FWIW, the AARoads forum member whose post prompted my question, now is less convinced than his original post several years ago that there are additional QC 138 segments we should add. So I would leave this alone for now.

BTW, he also notes that a connecting road under construction between the La Tabatiere and Tete-a-la-Baleine QC 138 segments will change their routing to bypass the towns in addition to merging the two segments. No change needed at this time.

=====

Other comments I agree with, except on the A-25/A-440/QC 125 interchange, which is giving me a headache and I'll curl back to it later.

[yakra]

MteeBaieST?

This sounds good.

I'll change 4 back to A-20(63).

This should stay 4, as it appears to be an "own-designation" exit number, following the adjacent 1 and 2.
Compare NY NY440 10

As for other examples, I'm scratching my head on that one.

Are you referring to when I said...?

Add a point in the middle of the rectangle bounded by Rue Notre-Dame Centre, Rue Saint-Georges, Rue Royale, Rue Saint-Roch -- leads to A-40 exit 199 and helps clarify the routing in the area a bit.

I'm suggesting adding a new point here.

The one that comes to mind is QC 116 near Montreal, which has its own exit numbers, including on a segment concurrent with QC 112.

The QC116-based exits, with the plain numbers-only waypoint labels, are definitely fine.
As for the A-20 concurrency & its labels and how well that adheres to the letter of the manual, well... that's a discussion that can get deeply bogged down mighty quick. If those labels are questionable, then I've also flagged similarly questionable labels on MA140, and not changed them. I'm OK with walking away whistling casually.

the A-25/A-440/QC 125 interchange, which is giving me a headache

Yup.

[oscar]

-- Any comments on how I dealt with the messy A-25/A-440/QC 125 interchange in Laval?

I made sure to have a look in the HDX before reading the rest of your thoughts, to see what solution I'd lean towards without any pre-bias.
Aye Caramba.

Moving the waypoint to the A-25/QC 125 intersection from the middle of the A-25/A-440 triangle better reflects that the interchange serves QC 125 and not just A-25 and A-440, with direct ramp connections from QC 125 to A-25.

My first thoughts were that this might lend itself to a 1PPI solution. The triangle, along with the QC125 at-grade junction thrown into the mix.

But I'm not sure about routing QC 125 via A-440(34),

What struck me was the angle of the final segment of A-440. Judging by a numbered Exit 34 being where it is (I'm not checking GMSV right now) it looks like A-440 heads south on the west leg, from the A-25 split at the N end of the triangle. The route trace looks awkward, but in a sense checks out in that A-440 continues to A-25, and that's where A-25's point is.

when northbound QC 125 traffic goes east of that point (southbound QC 125 traffic has to go through that point, to connect to the QC 125/A-440 concurrency).

Nonetheless I think 1PPI may be the cleanest -- a "correct by way of vagueness" -- solution here.
With the Masson/Marcel-Villeneuve intersection right on one edge of the central triangle, you could let this weight your judgment of where the triangle's center is.
I took the I-97@7 example to extremes, and considered where traffic would go if the thru lanes weren't available, if one were to exit & re-enter the highway. The inside-the-triangle, left-turn-ish movements would pass thru that intersection. The outside-the-triangle, right-turn-ish movements would be a little ways out on each leg.

There's another situation like this in TX, with a central ramp triangle, and a junction of at-grade routes right on one of its legs.
The main difference is there's not the added complication of a surface route joining/leaving a freeway via some funky ramps that QC has. (But there's still access between all the things because frontage roads.)
Here's how I handled it:

ISTR pondering whether to slide the point a bit more SSW, and could just as easily have done so.

One difference is, in the Texas example there appear to be no separately-numbered interchanges within that overall interchange, though there are 293A and 293B ramps. QC 440 has separate, signed 34 (Av. Marcel-Villenueve, etc.) and 35 (A-25 southbound -- thru lanes connect to A-25 northbound) interchanges, with 34 looking like an interchange-within-an-interchange. 34 covers connections between QC 125 and A-440, and WB A-440 from SB A-25. 35 covers connections from SB QC 125 to A-25, with a direct ramp to NB A-25 and the funky ramp to SB A-25, in addition to the connections to A-25 from EB A-440. Waypoint 35 (A-440/QC 125)/17 (A-25) is placed where QC 125 passes under A-25, and near the direct ramp from QC 125 to NB A-25, which is also not a bad placement for the ramp triangle between A-25 and A-440.

This situation, ISTM, points to a 2PPI solution, if only to account for the two signed exit numbers 34 and 35 within the overall interchange.

I plan to pull in route file changes tonight that will implement your other comments not previously addressed, but no changes to the A-25/A-440/QC 125 complex. After they're all pulled in and I've double-checked them for NMP and Datacheck errors, I'll activate canqc tomorrow (quite by coincidence, the day I'm let out of my cage, as the stay-at-home order for northern Virginia expires).

[Jim]

Thanks Oscar and everyone else who got this ready to go.  I've pulled in the activation and I'm running an extra site update now.