HSL DateTime: Time (interval) class

[jjergus]

This is a small stand-alone initial part of the DateTime library (the HH\Lib\Time class, which doesn't depend on anything else, and an associated test).

I'd like to get this reviewed before I spend too much time working on the remaining parts. Here I am mainly looking for feedback on:

• Any high-level/general feedback about the proposed API/implementation
  • by "general" I mean anything that I would need to take into account when working on the remaining parts; feedback about specific details like "the toString method is terrible" is welcome as well, but not as important -- specific details can always be changed later, but if anything looks like it's going in some bad general direction, I'd like to know before I go any further in that bad direction
• Is the amount of documentation and test coverage here too little/just about right/too much for an hsl-experimental library?
• Should I separate the rest of the library into similar smaller parts like this, to be reviewed and committed separately, or is it better if I prepare everything and have it reviewed all at once?
  • You can always look at https://github.com/jjergus/hsl-experimental/tree/dt4/src/datetime to see how anything works in context (there's also README.md there explaining most of the design decisions).
• Do we need to decouple hsl-experimental from FB WWW before I start committing these? I think we probably don't want to have these in the HH\Lib namespace in FB WWW, but either of these seems reasonable to me:
  • put these in HH\Lib\Experimental namespace and commit to WWW
  • keep these in HH\Lib but decouple hsl-experimental from WWW first
  • or even both: decouple hsl-experimental from WWW and also move these to HH\Lib\Experimental
  • having these in FB WWW would mean people could start using them, which has disadvantages (hard to change things later), but also advantages (much better chance to get useful contributions from FB engineers? I expect the first person/team that wants to use it would effectively become the main maintainer, which could be better than it effectively becoming unmaintained when I stop working on it)

[fredemmott]

I'll look more thoroughly through soon, but before I forget:

keep these in HH\Lib but decouple hsl-experimental from WWW first

Raw HH\Lib\ should really be preserved for built-ins only. It is a lot of work and pain to move stuff from external to built-in without changing the namespace, and essentially forces a minimum 3-week delay to the eventual merge process, to accommodate our need to be able to test/run on older versions of HHVM for a while.

[lexidor]

Note to self, have a look at dt4 shortly. The API has changed since dt2.

[jjergus]

• moved to ...\_Private\_DateTime
• added fromParts
• removed optional arguments
• added __debugInfo() (but I kept __toString() as well, it seems more generally useful to me)