Special rendering for intermittent waterways - fixes #805 #1000
Conversation
|
Do we want to render intemittent waterways different from wadis? |
|
I considered treating wadis as intermittent rivers, but I am not sure is it a good idea. Also, according to http://en.wikipedia.org/wiki/Wadi wadi is rather term for an entire valley - not just intermittent waterway. |
|
The problem is if wadi is understood as a synonym for intermittent waterway it should be rendered the same. If it is understood differently it should probably not be rendered implying a waterway at all since it can be permanently dry. Rendering it differently but still waterway-like is the worst solution IMO. I would probably change the dashing for rivers to make sure the dashes are always at least as long as wide. |
| @@ -133,6 +138,19 @@ | |||
| [zoom >= 18] { bridgecasing/line-width: 13; } | |||
| } | |||
| } | |||
| [intermittent = 'yes']{ | |||
There was a problem hiding this comment.
I think in our coding style, we use a space before {.
|
I agree that wadi and intermittent rendering are too similar in this proposal. Either we need to render them such that the difference can easily be detected, or we need to render them the same. I would prefer the latter: I checked with Overpass Turbo, and it seems that waterway=wadi is currently being used in the same way as intermittent=yes. In places where intermittent is used, is it also used for riverbeds that have been dry for years? |
|
The white with dashed outline could be misunderstood as a road. IMHO, the dashed blue line works nicely and should be working across all widths. "Ditch" kind of implies "intermittent" already. See the wiki http://wiki.openstreetmap.org/wiki/Tag:waterway%3Dditch They may contain little water or even be dry most of the year. |
OK, I will merge wadi and river/canal definitions. I will also fix coding style.
I am not aware about this kind of geographical features in Poland.
Yes, I am not expecting that something like exists in data - it was mainly sanity check to avoid some weird bugs.
I admit that I missed it (in my mapping I was adding intermittent=yes but never intermittent=no to ditches).
I think that it is unlikely.
This examples are already with wider ditches (I branched from the last trunk version). |
|
Coding style fixed, river/canal and wadi styles merged. |
|
For ditches - explicit intermittent is IMHO better, implied tags that are not super obvious lead to problems (and I modified wiki page about ditch with recommendation to tag this way). |
Yes, definitely. Funnily, the German wiki page already had the recommendation to use "intermittent=yes". |
|
For river, and perhaps even for stream, I would go for a bit longer dashes. Maybe the same length/width proportions as for drains? I think it would make intermittent rivers and intermittent streams look more similar. Apart from that, code and resulting rendering looks good! |
|
Oh, and:
I wouldn't worry about this case. |
It turned to be a good idea for rivers.
Still, I think that streams are now OK. |
|
Note: once changes stop, I will test various edge cases to ensure that nothing is broken and rebase everything into a one commit. Also, is it possible to maintain consistent dashes across (meta)tiles?
|
Hmm, now I see the result, I'm no longer sure if it looks nice. What are the opinions of others?
pI'm not sure if there is a way to do that, maybe @gravitystorm or @pnorman know? |
|
I used short dashes as I was sure that longer will look like series of lakes rather than a linear object (it was worse with round caps). I like both versions, though metatile glitches seems to be less iritating for a short dash version. |
|
At the highest zoom levels where rivers are drawn very thick it would probably be good to style them in a way that works well together with the style of intermittent water areas (see #996). One option for the high zooms would be to use a style similar to highway=proposed, just without the white parts, i.e. a thin non-dashed casing and a dashed interior, all in the normal water color. |
Yes, but currently it is unknown how #996 will be solved. But with this proposal intermittent waterways should not be more troublesome than a normal ones unaffected by this change.
I am unsure how it may be in CartoCSS. Standard method will not work, as it is based on painting one color wider tha placing narrower dashes on it ("Another common railroad line style is similar" from https://www.mapbox.com/tilemill/docs/guides/styling-lines/ ). Also, such symbol change depending on a z-level may be confusing. |
You can probably do that with line-offset (two thin solid lines with offset and one thick dashed line without). It will of course look somewhat ugly when several waterways meet.
I don't think this would be a problem. The thinner the whole river line gets when you zoom out the thinner the outer lines will be as well and at one point you simply leave them away completely leaving only the dashed center line. |
Turn off clip |
|
@pnorman - thanks! What fixed one bad consequence of long dashes, but I found another problem with long dashes.
so I roll back to a short dash version
I think that in this version "thin non-dashed casing and a dashed interior" is not needed. |
|
Still needs rebasing and retesting, I plan on do this tomorrow (maybe somebody has some comments). |
Minor issues will always exist with display of lines, and especially dashed lines. I don't know if that needs to be a reason to "roll back" or refrain from using them. Personally, I think the very short dash, is to much like some kind of "steps"... It doesn't appear to work cartographically. I think the medium or long dashes you showed in these images appear best. Alternative is the dashed outline you showed, but that one could be confused with a motorway tunnel considering the close resemblance in color, although the general irregularly bended nature of it and no connections with other highways, will likely avoid this confusion. |
|
The problem is that these problem appeared on many curved segments of rivers, it was not something rare. But I am not strongly against long dashes. Is there anybody else who have an opinion on long dash vs short dash issue? |
|
What about blue dashes on some grayish/brownish background, or something like that? |
|
It was not working well with white casing for streams and ditches (that is why for intermittent small waterways casing is also dashed) - it was quite messy except on high zoom levels. And I think that the doing it only for rivers may be confusing. |
|
Maybe you could also reduce the rendered width of the major waterways a bit on some zoom levels. It is not entirely clear to me from the images you posted what the actual rendered width at each zoom level is, but it appears quite wide at some. Reducing it might also reduce the visibility of possible artefacts. |
|
@mkoniecz Could you rebase this? |
|
@mkoniecz Can you rebase this? |
|
Yes, I will do it tomorrow. |
Includes change to wadi rendering.
|
Rebased, now I am testing whatever everything works properly. |
|
Probably too late for this release, but according to my tests everything works properly. |
Note: there are still some issues.
(1) - in a rare situation - intermittent waterway with bridge=*, connecting with not intermittent waterway ordering of rendering may be wrong, leading to a small glitch. It is fixable by ordering waterways in SQL querry - but is it worth additional complexity and performance hit to solve something that never appears in data? Or in the worst case - is really rare? (based on math1985's comment)(2) - is it possible to maintain consistent dashes across (meta)tiles? See the first image. (thanks @pnorman!)(3) - is it necessary to modify rendering of wadis? And (changed) derelict canals? Both use dashes, but wider and rounded ones. See examples at http://www.openstreetmap.org/way/81918899 and http://www.openstreetmap.org/way/220089162#map=16/31.0738/34.5945 (wadi rendering is changed, for derelict canals I opened dry canals are displayed as blue dashes #1003)(4) - is it a good idea to assume that ditch is intermittent ? (no)Examples of a new rendering on an example of http://overpass-turbo.eu/s/5ec (+some minor modifications of OSM data)
Minor waterways:
Major waterways:
