Author: Ldp
[Added to the original trac issue at 10.47pm, Tuesday, 2nd March 2010]

This is by design, to prevent completely massive shields from popping up on the map. Ideally, we would like to render a shield for each of the refs in such a composite ref, but this is not yet easy to do in mapnik.

Author: NE2
[Added to the original trac issue at 2.58am, Wednesday, 3rd March 2010]

How simple would it be to "cut" the ref at a semicolon so it displays, even if not all the routes are shown?

The single routes could be solved by making common abbreviations like Alternate->Alt and Business->Bus (yes, signs often abbreviate it Bus, despite possible confusion with a bus roadway, especially by analogy with Truck). The limit (currently 8) would have to be upped slightly (to 10) in these specific cases.

Author: Ldp
[Added to the original trac issue at 9.13pm, Thursday, 4th March 2010]

Replying to [comment:2 NE2]:

How simple would it be to "cut" the ref at a semicolon so it displays, even if not all the routes are shown?

Relatively simple, although if the total compound string length is under the max, and you actually do want to show both in a single shield, things get complicated again. Some refs, like 'A4;A29' are actually short enough to render already.

The single routes could be solved by making common abbreviations like Alternate->Alt and Business->Bus (yes, signs often abbreviate it Bus, despite possible confusion with a bus roadway, especially by analogy with Truck). The limit (currently 8) would have to be upped slightly (to 10) in these specific cases.

Current OSM practice is to not abbreviate anything, at least for street names. I don't think it would be good to loose the unambiguous 'Business' for 'Bus' in this case.

Author: NE2
[Added to the original trac issue at 11.08am, Friday, 5th March 2010]

Replying to [comment:3 Ldp]:

Replying to [comment:2 NE2]:

How simple would it be to "cut" the ref at a semicolon so it displays, even if not all the routes are shown?

Relatively simple, although if the total compound string length is under the max, and you actually do want to show both in a single shield, things get complicated again. Some refs, like 'A4;A29' are actually short enough to render already.

My idea was to only cut it if it would otherwise not show. In other words, find the last semicolon in or before position 9, and render only what's before it.

The single routes could be solved by making common abbreviations like Alternate->Alt and Business->Bus (yes, signs often abbreviate it Bus, despite possible confusion with a bus roadway, especially by analogy with Truck). The limit (currently 8) would have to be upped slightly (to 10) in these specific cases.

Current OSM practice is to not abbreviate anything, at least for street names. I don't think it would be good to loose the unambiguous 'Business' for 'Bus' in this case.

Oh, I mean having the renderer do the abbreviation.

Author: NE2
[Added to the original trac issue at 10.36pm, Thursday, 21st April 2011]

Bumping this up to critical. Sorry if this is a bad thing to do, but this is a major issue in countries with overlaps. As two examples, these long sections of freeway get no labels:
http://www.openstreetmap.org/?lat=32.4068&lon=-88.5502&zoom=13&layers=M ref=I 20;I 59
http://www.openstreetmap.org/?lat=41.6321&lon=-108.3838&zoom=13&layers=M ref=I 80;US 30

Some time ago, semicolons were changed to be parsed as carriage returns in highway=motorway_junction name=* tags. This might be the easiest solution: create a multiline label and expand the rounded rectangle to fit. (It wouldn't fix 'bannered' routes like Alternate and Business, but those can be handled separately.)

Author: Ldp
[Added to the original trac issue at 10.49pm, Thursday, 21st April 2011]

Then again, if you remove those unneeded spaces from the refs, they're suddenly 2 characters shorter and might render. ;)

Not a fan of multiline refs the way you describe. I'd rather get them all rendered as single shields. I once tested that, but mapnik just kept trying to put the shields for the 2nd ref at the exact same locations as the already rendered 1st ref. I should try again with current mapnik versions, or nag the authors a bit more.

Author: NE2
[Added to the original trac issue at 1.13am, Friday, 22nd April 2011]

Replying to [comment:6 Ldp]:

Then again, if you remove those unneeded spaces from the refs, they're suddenly 2 characters shorter and might render. ;)

The spaces are part of the standard, and are necessary where alphanumeric designations are used (how would anyone know that CRAP is County Road AP?) And if one were to remove the spaces just for the sake of rendering, these would just barely fit, and longer ones still wouldn't.

Author: Ldp
[Added to the original trac issue at 8.07am, Friday, 22nd April 2011]

Replying to [comment:7 NE2]:

The spaces are part of the standard,

OSM or US ?

and are necessary where alphanumeric designations are used (how would anyone know that CRAP is County Road AP?) And if one were to remove the spaces just for the sake of rendering, these would just barely fit, and longer ones still wouldn't.

I know it will just move the problem to longer compound refs. Mind the smiley.

It wasn't possible in the past to render the constituent refs as separate shields, but as I said, I'll try it again soon.

Author: NE2
[Added to the original trac issue at 1.37am, Tuesday, 26th April 2011]

I just looked through an old Rand McNally atlas and it uses multiline shields for overlaps. It may not be as good as separate shields, but we shouldn't let the perfect be the enemy of the good.

Author: rw__
[Added to the original trac issue at 4.49pm, Monday, 5th September 2011]

Not a defect. Not Critical. Other solutions available.

Author: rw__
[Added to the original trac issue at 4.50pm, Monday, 5th September 2011]

Closed.