Avoid a new line break at the hyphen in "Co-op" in labels

[pkoby]

Expected behavior

As such a short word, it's not expected to see "Co-op" broken into two lines, especially with the weak vowel in the second unstressed syllable. I think it should be kept as one word, if an exception could be made to a rule. I couldn't find where that might happen in the code, or I would have looked into it myself.

Actual behavior

It is split after the hyphen in some instances.

Screenshots with links illustrating the problem

See https://www.openstreetmap.org/#map=18/40.053459/-75.171885

Also here: https://www.openstreetmap.org/node/11260100599#map=19/40.046520/-75.195960
And here: https://www.openstreetmap.org/node/2760113374#map=19/40.075227/-75.205517

[imagico]

Implementing line breaks correctly across multiple languages is a hard problem, especially since we use the same unicode characters in different language where different line breaking conventions apply.

The most reliably way to ensure correct line breaks is for mappers to differentiate between hyphens allowing a line break and those which do not in tagging. Quoting from the Unicode Line Break Algorithm - which Mapnik uses by default:

The rules for treating hyphens in line breaking vary by language. In many instances, these rules are not supported as such in the algorithm, but the correct appearance can be realized by using a non-breaking hyphen.

If there is no consensus to document this differentiation in tagging of names (and you could indeed argue against that) other solutions would be to

• maintain a dictionary of words in which hyphen should not allow line breaks and substitute such hyphens with non-breaking hyphens in data processing.
• maintain a set of per-language rules that substitute hyphens with non-breaking hyphens based on context (like the short word aspect here). This requires knowledge of the language of the name tag (which we could use also otherwise - see What about Han unification? #2208).

[pkoby]

I had no idea that a non-breaking hyphen was an option (or even existed), and I have no problem using it for my uses. However, it seems like a non-breaking hyphen might not be recognized as a hyphen in searches. Thoughts on that?

[imagico]

As i said - there are valid arguments against differentiating that in tagging. I am not aware this has ever been discussed more broadly in the mapper community.