Yeah, I have tried to generate cycle routing in many engines and they all feel like beta versions. I wish there was an easy country-based solution to routing that didn't require everyone to jump through hoops, but I also believe OSM needs to be consistent not just at a country level but worldwide level to remain useful and competitive. Adding inconsistent data now will inevitably lead to maintenance for decades to come. I think this is the view held by most mappers with allowance for discrete exceptions. I guess I digress a little here, but since you mentioned motivations, I certainly err on the side of correct (whatever that means) in long-term.

You mention a trend to remove `bicycle=yes`. One thing I will disagree with is describing removing `bicycle=yes` as a trend rather than the trend that people have been *adding* `bicycle=yes` without broader discussion, without referring to any OSM guidelines and generally using personal experience. The few mappers I have asked either mentioned a specific router they had issues with, did not yet understand the traffic laws or just didn't really understand what access tags mean. In any case, there was never any broader discussion to fix such broad router issues with local OSM data changes. This sort of `bicycle=yes` tagging on sidewalks was literally everywhere from major trunk sidewalks to random living street sidewalks to sidewalks literally next to cycling infrastructure. So I did remove all of it fairly indiscriminately, although I did look at each place briefly and I'm more or less personally familiar with all of them. That said, places like bridges or your example are certainly one of the most likely candidates to be an exception.

I would be curious to see how mappers deal with what routers do in all the other countries that have similar rules to Latvia. In fact, Latvia is one of the last "developed" countries to actually add the explicit "no bicycles on sidewalks except..." and make an "official statement".[1] Many other countries have had this codified for a long time. And terrible routing is certainly not exclusive to Latvia.[2] "Avoid bicycles on sidewalks" is pretty much the default scoring for all routers. It's just that we have so little cycling infrastructure that it's incredibly obvious. And it's not even sidewalks, routers just bias roads a lot.[3] And given that this is a systemic problem with routers, I don't see how we can solve it on OSM end. Like, if every other sidewalk we map needs to be an exception, then clearly something else is fundamentally wrong.

The "solution" to connect the cycleway to the highway is technically how it's usually done, yes, at the end of cycleways that have to merge into traffic. It would be incredibly stupid here, but that is the legal "primary" route with sidewalk being the alternative. Although I should note in this particular example, I don't think any of the default routers will score the route to actually use the trunk road over other sidewalk "alternatives" and it's already connected to the trunk at the crossing. Even cycle.travel, which is arguably the best cycle router, does crazy stuff in this location (although it does actually route along the trunk road in this section if you put the points apart).

[1] https://www.lsm.lv/raksts/dzive--stils/motori/lukstins-ietve-ir-paredzeta-tikai-un-vienigi-gajejiem-velosipediem-iznemuma-gadijumos.a127199/
[2] osm.org/directions?engine=fossgis_valhalla_bicycle&route=60.384%252C23.127111%253B60.383931%252C23.125663
[3] osm.org/directions?engine=fossgis_osrm_bicycle&route=57.313513%252C24.41389%253B57.311653%252C24.412565

Hello,

What is this road way/1425453312 supposed to be here? It's not on the aerial. If you have surveyed this as a new road, then it needs an actual classification.

Legally, cyclists are allowed on sidewalks as an exception, not the rule. Bicycles are to use cycling infrastructure or roads. In other words, in Latvia on sidewalks, it's `bicycle=no` (like most countries). Now, the exceptions to use sidewalks in Latvia are very broad - traffic, weather, obstacles, etc. And enforcement is non-existent. Plus, it's culturally so embedded that bicycles use sidewalks that hardly anyone even knows the actual traffic laws. So to compromise, we don't actually put `bicycle=no` on sidewalks, we just leave it with the default value and let routers (profiles) decide if they want to follow the letter of the law or be more loose with restrictions (which most do). But putting `bicycle=yes` implies that it's no longer the default restrictive traffic rules, but bikes are specifically explicitly allowed by some sign or something. That's just not the case for 99% of sidewalks and is incorrect data.

But I guess that's the problem with incomplete car-centric routers, isn't it. Do we map for the routers or do we map the actual restrictions? Because if we choose to map for the routers, then we need to add `bicycle=yes` to every sidewalk across the entire Latvia. Because we can't just subjectively cherry-pick some of them to "fix" routing for specific routers in specific locations. This would fail OSM subjectivity and verifiability principles. And we would be doing something weird just for Latvia. Other countries that have similar traffic laws don't do this (as far as I know).

Your example is a great example of what happens with automated programs when cycling infrastructure in Latvia sucks, while laws do not represent reality. So, legally, you should attempt to cycle on the road here and if that's dangerous or impossible, only then use the sidewalk and then cycle at slow speeds, give way to all pedestrians (which frankly are here always), slow at intersections, etc. Is this stupid and should this logically be at least a shared cycleway? Sure. But that's how it's legally defined and that's how any accidents here will be treated. So, yes, I call `bicycle=yes` "bogus" and that's putting it mildly, because it's legally `bicycle=no` with a long list of exceptions here regardless how people actually use it.

Are current laws stupid? Yes. Is infrastructure bad? Yes. Are router profiles poor? Yes. But I don't think the solution is to massage OSM data to fit them for something so broad. I can understand some exceptions in some cases. But here we are talking "sidewalks" as a concept throughout all of Latvia. I personally generally describe sidewalks in Latvia akin to `bicycle=permissive` in practice and I would agree to make this a country-level default access for routers even if it's not technically correct. If I recall right, I mentioned this before although we never got around to finishing the country defaults table.

Now, all that said, if multiple mappers discuss, argument and agree to mark certain road segments like this with `bicycle=yes`, then that's fine by me. But this *has* to be discussed and recorded, we can't just have everyone individually tagging subjectively to improve their favourite router.

It looks like you have not fixed the issues, but are continuing to edit elsewhere and have not replied to any of the recent changeset comments. Because this changeset breaks navigation, I have reverted it for now.

Similarly, these were both added at the same location node/8709526415 and node/8709526379 .

In this changeset, you changed an apartment building into residential landuse. I am not sure what you were trying to change here, but I have reverted this for now as it seems to be a building according to aerial and cadaster data.

Sveiki,

Nosaukumus tulkot nevajadzētu, ja tie nav oficiāli nosaukumi. OSM nenorāda tulkojumus, bet jau esošus nosaukumus (kas var būt tulkoti, bet tos mēs netulkojam paši). Piemēram, tūrisma mājas lapā vai uz informācijas dēļa vai kādā oficiālā datubāzē. Lielākajai daļai vietu nosaukumu citās valodās nav. Citādi sanāk patvaļīgi tulkojumi, kas var arī nebūt pareizi. Piemēram "dzirnezers" nav "lake" un "tautas nams" nav "community house" (ja vien tas nav šādi nepareizi tulkots oficiāli) utt. Arī īpašvārdi netiek burtiski tulkoti piemēram "Baznīcas ceļš", kas tad būtu "Baznīcas road" vai "Baznicas road". Skatoties piemēram https://kandava.lv/en tulkots tur nav gandrīz nekas. Šādus aprakstošus tulkojumus var gan pielikt `description:en` tegā.

You added a fence here instead of what I assume is a driveway. This breaks routing for the road and footway that are now crossing a fence, so please be more careful.

I have reverted this edit because it added a large amount of invalid fords breaking routing along all these roads. Please review your edits before adding so many features that don't exist.

Please actually fix issues with the roads that you add using Rapid, which is your responsibility if using semi-automated tools. You left the road with an incorrect layer value, overlapping buildings, overlapping a wall, impossible geometry, misaligned.

Can you please elaborate on your edits here? What is this ditch that flows uphill and connects to trenches way/1423368941 ? Why are there ditch culvert between trenches like way/1423368945 , what is here under the tracks and road ? Some of trenches you drew literally cross a highway like way/1423368926 .

Please fix these features.

Sveiki,

Latvijā mazciemus OSM apzīmē ar `place=hamlet` (nevis ar `place=village`, kas ir "pilnie" ciemi ar VZD robežām). Es izlaboju. Ņem vērā, ka Wikidata un citi avoti var būt novecojuši/nepareizi. Pilni VZD dati ir pieejami https://data.gov.lv/dati/lv/dataset/varis-atvertie-dati .

Hello,

How do you know in this changeset and others that these are fords, e.g. node/13094512758 or node/13094512977 or node/13094512859 and others ? They do not look like fords and it's very unlikely larger rural roads have fords like this.

This breaks navigation along these roads, so please check all of these locations promptly.

Sveiki,

No kāda avota tu šeit iezīmēji takas?

Piemēram way/1421214349 (kas dabā neeksistē ar apgāztiem kokiem) vai way/1421214358 (kas iezīmēta pāri sētai slēgtā veikala teritorijā) vai way/1421214361 (kas daudz gadus jau aizaugusi) vai way/1421214350 (kas iezīmēta caur sētu privātīpašumā) ?

Šajā piemēra no novecojuša ortofoto izmaiņu sanāca maz, bet es to pieminu, jo var sanākt arī krietni vairāk kādā citā vietā. Labot ir parasti zināmi grūtāk nekā zīmēt jaunu, it īpaši, ja ortofoto nav pieejams un jāzīmē no mapillary/video.

Platformu tiešām es nebiju vēl paspējis pārzīmēt, jo nebiju vēl nobraucis pa to (tā ir īsāka). Paldies, ka sabīdīji.

Tas, ka parku iezīmē par laukumu ir ļoti labi. Bet runa ir par to, ka tavā izmaiņā ir izdzēsti meža tegi. Bet pats parks izveidots no nesavienotiem gabaliem, kas neietver ielas, kas ir parka teritorijā, bet ietver privātīpašumus, kas nav parka teritorijā. Un atstājot neizdzēstu parka punktu kopā ar laukumu sanāk datu duplicēšana, kas kartēs un pārskatos šeit rādīs divus parkus. Es centos saprast tavas izmaiņas rediģēšanas mērķi, lai saprastu, ko izlabot.

Lūdzu neveido parku šādi kā multipoligonu no fiziskiem meža gabaliem. Vai arī tad šeit jāveido multipoligons are jauniem posmiem un tas viss jādala un jāsavieno pa atbilstošām līnijām.

Es parka teritoriju pārzīmēju ņemot vērā privātīpašumus. way/1421757069

Lūdzu ņem vērā, ka, tā kā LVM ortofoto ir vecs un pārbūvēto staciju tajā nav, tad ar to kartēšanu vajadzētu uzmanīgi, ja nav personīgi apsekots un ja iepriekš kāds jau kartējis jaunākas izmaiņas. Šajā izmaiņā ir pievienoti pāris demontēti soliņi un nocirsti koki stacijas ēkas ziemeļos. https://www.mapillary.com/app/?pKey=1275005580720821&focus=photo Es šeit izlaboju.

Sveiki,

Vai varētu precizēt, kas šajā izmaiņā ir mēģināts izdarīt? Ar šo izmaiņu mežs ir izdzēsts, bet nesavienoti gabali ir pārveidoti par parku. Tajā arī iekļautas privātīpašuma teritorijas bet nav iekļautas pastaigu ielas. Un parks arī tagad ir divreiz kartēts, jo bija jau punkts node/10796729145.

Hi,

While I personally don't know the situation here and was just fixing invalid values, I should note that A/P/V to trunk/primary/secondary is the recommendation and starting point, but it is not necessarily the final decision. OSM classification is based on function and logical road network, while A/P/V are just the codes used by LVC that happen to mostly match. But frequently in Latvia P and V roads are not actually used in real-life how classification would suggest - a lot of them were assigned historically and internally and LVC doesn't really make major changes to codes even when real-life situation changes. So someone with better local knowledge might retag them higher and lower to build a more logical hierarchical network for navigation. I have myself previously surveyed roads and retagged them seeing how in real-life the P/V/x values were old, outdated, rerouted, etc. If this P66 is a gravel/compacted road and in practice all people use P68, then I can see how this might actually be secondary or even tertiary compared to P68. Based on OSM road network principles, you generally want road classification to decrease as the roads branch away and people don't use it as the main way for travel/navigation. Anyway, I'm just noting that it's possible this to be secondary and P, but as I said I don't know the local situation. You would have to ask the original editor who adjusted it. I didn't want to change the classification here because I would just be guessing and it looked okay at a quick glance.

Hello,

Could you please clarify about this point here node/12879605908 . You have mapped both a bridge and a ford. It's very unusual for it to be like this. I am assuming this was a mistake and it's either a bridge or a ford (or perhaps a ditch culvert)?

Thanks