Slow time with natural language

[lode]

Adds an option for slow time using natural language. This is roughly based on the idea of @Cj-Malone in #6.

I'm unsure about the locale changes with translations from format.js as I had some trouble building these translations in the .po files. Please let me know if this is not done in a correct way!

[da2x]

I’ll need some time to think about the possible localization issues this may cause. Help in reviewing this would be quite welcome.

[brandl-muc]

Code duplication, would be better to create a function that does the 0-23 to 1-12 conversion.

[brandl-muc]

Looks good, it would be nice to add tests though once #33 is finished.

[brandl-muc]

I’ll need some time to think about the possible localization issues this may cause. Help in reviewing this would be quite welcome.

Can you elaborate on the possible issues? Are you afraid that some languages cannot be represented with the given solution?

[da2x]

I really don’t know what the localization issues are, but I expect there to be many.

[brandl-muc]

I guess you are referring to something like this: How To Tell Time In Russian
Although this page suggests that it would actually still work: Time in Russian

I guess the only way to play safe would be to have a localizable string for all 48 combinations in the 12 hour system. Would that be a better solution?

[da2x]

Danish and French use vigesimal numbers (base 20). I'm assuming they also more often round to 20 and 40 instead of quarters when telling time "naturally".

It's regional stuff like that I'm unsure about. I was hoping to find a more comprehensible resource covering common time expression internationalization issues, but few cover natural language (at least with English translations).

[brandl-muc]

I guest most peoples today really use a decimal system. I didn't and don't know about the Danish, but the French really only have irregular names for several multiples of ten - but also only those greater than 60.The clock should therefore be no problem for them.
On the other hand halving seems natural, so halving an hour and then again halving that seems agnostic of the counting system.

But to actually make progress I only see three possibilities:

• decline the PR
• accept the proposed system and wait for someone to complain that their language cannot be represented (probably unlikely to happen)
• go for the 48 individual strings given that halving the half should be a universal thing.
• OK a 4th: use the original proposal of Fuzzy clock #6, maybe replacing quarter with fifteen for the sake of consistency.

We sure can wait some more but I doubt this will shed more light on the subject.

[brandl-muc]

Trying to revive this one too. I'll try to argue for the 48 string approach. If it still doesn't sound like a good idea I would ask for declining this PR.

There are a few resources on the topic:

• Clock-time expressions in the colloquial language: some suggested universals
  They have an Euro-centric view which they admit. In the comments it is noted that Arabic may have thirds, however that seems like an additional option, as this blog teaches quarters and halves only.
  The article is based on a German master thesis on the subject. I could try to get a copy, however I doubt that it will reveal a lot more.
• This Quora thread collected many different forms of telling the time, having full and half hours seems universal.
  Many languages have quarters as well, and those that don't seem to simply use something along "15 past" and "45 past" / "15 to".
  As an extreme, Cantonese uses pentaminutes, so it would be "3 past" and "15 past". Quarters might not stand out that much here but there is definitely a natural colloquial form to designate that time point.

This strongly suggests that the basic concept of displaying the slow time based on quarters is sound. However I don't think that the code as it stands does the trick since languages might decline numbers differently with their counterparts of "past" and "to". (Russian in fact does just that)

Therefore, I would offer to adopt this to a 48 string system in case you would then accept it.