In fact, just follow this wiki and we will be both happy:
osm.wiki/Bicycle

Please follow this as I though we both agreed on that:
When there is a physical separation (curb, different level, grass or step) between road and cycleway: use a separate way for cycling.
When there is no separation 9same level, only paint on road): use cycleway:lane on road way. And you must specify cycleway: left or cyclway:right, then cycleway:left:oneway = no
or cycleway:right:oneway = no when cycling path is not oneway

The track must stay separated from road here until Rue Cardinal, since there is a physical separation between track and road (curb). Use cycleway lane when the cycleway is on the same level and inside the street (no physical separation).

should be fixed now. I will keep the sidewalk=separate with both left and right subtags for compatibility.

Well, unfortunatly, some routing engines parses only separate and not separate:right and separate:left so we still need to keep the general separate tag.

Would it be fine to use:
sidewalk=separate
sidewalk:left=no
sidewalk:right=separate

OK, I will try to fix these

Every routing engine has a different process to parse tags. In OSRM, we can specify use_sidepath as foot=no, so everything works. And in our analysis, we use sidewalk=no, right, left or both to characterize pedestrian security and quality. I am open for suggestion to change this behavior, but it would invlolve lots of works on our side to modifiy the paradigm. Who decided to trigger an error in Osmose for that specific issue?

sidewalk=both specifies that there are sidewalks on this road. use_sidepath just tell routing engine to use the sidewalk way instead of the street way. This should not trigger a problem, since the sidewalk tag is used to characterize streets. These are two different informations. When we compare neighbourhoods for pedestrian network quality and security, we use the sidewalk tag to find the percentage of streets with no sidewalk, at least one sidewalk or two sidewalks. When we use a routing engine, we use the use_sidepath tag instead.

Fixed!

You can see the replacements in progress in my other changesets: changeset/80481444

These residential zones are outdated. We are replacing them with more precise zones that do not include roads and parks.

Thanks! Do you have a tool to find such typos?

FIxed, I replaced them by student:count

Thanks for the info, I looked in the official wiki but did not find any tag for this. I will use one of these.

Bonjour!

Cette piste cyclable est-elle complétée? Attention de bien la connecter au reste du réseau routier. Elle était connectée seulement aux polygones, qui ne sont pas "routables". Merci et bonne journée!

Désolé, je voulais simplement créer un cercle parfait... C'est bizarre que iD laisse faire le merge si des relations sont dans les ways impliqués...

OK, so finally, I removed the illegal crossing since it is written on the CN website that it is always illegal to cross if there is no official rail crossing, but this will make routing engines very unrealistic in these neighbourhoods.

If you are talking about the Maricourt Boulevard and the Kimber Boulevard, there is no indication that it is forbidden to cross the track there and there is no fence at all. Everybody just crosses the track there wihtout any problem whatsoever. I understand that this may be illegal, but most of the time, when the CN or CP doesn't want people to cross, they add very clear signs to alert pedestrians. I honestly don't know what to do here, since removing the paths will make the routing engine to return large detours which absolutely nobody will do in real life. I think we should at least make them access=permissible

What do you think?

I use an up to date version of id editor that I compile on my own machine, so that is why localhost is appearing. No problem here. For the google imagery, I did not used it since it is blacklisted in id editor and selecting it will just go back to none, but it appeared in the list and I don't know why it was chosen since I used Mapbox and ESRI. I will try to fix this in the code...

Hi!

I understand your concern, and I may have added too many paths to connect to nearest streets, but we need to make sure openstreetmap is suitable for realistic routing purposes, which is one of its main use these days (osrm, mapbox, tom tom, etc.), and for pedestrian routing, we need to add paths to connect parking aisles to roads even if there is no official foot paths, because otherwise, it would mean too many detours and that would render pedestrian routing completely irrealistic.

The same thing apply to bus lanes: when they are tagged on the same road segment, we cannot differenciate between congestion speeds for cars based on gps data with bus speeds which are faster on bus lanes.

I think it is a good compromise to create separate bus lanes segments. If we do not, we need to create complicated custom parser for osm data to recreate these bus lanes for routing.

As a side note, we, at the Chaire Mobilité, will complete and enhance openstreetmap data for the whole of montreal area in the next months/years to make it completely routable, including bus lanes, cycle lanes and pedestrian paths. For pedestrian paths, I would really like to use a custom tag (like "plausible pedestrian connector" for routing purpose only) but I did not see any official suggestion on this issue yet. If you have any suggestion, don’t hesitate to tell me!

Thank for your understanding!