Fix multiday events rendering bug

[antoinerousseau]

Fixes #36.

Inspired from SlavaKvak@a359db3

The main change is that slots and span are now rounded, the same way padding was already rounded. No Math.floor anymore.

[antoinerousseau]

I replaced the bug with another: the event is now properly split accross two rows in both en and fr cultures, but its length is wrong in the en culture: it is one day too short.

Test event starts on 2016-04-01 12:00:00 and ends on 2016-04-04 12:00:00

[antoinerousseau]

And @okcompute king of bug is still present, but in en culture only:

[antoinerousseau]

After some debugging, it seems that the problem comes from the rounding:

Math.round might have to replace Math.floor (not sure why these numbers are not already rounded since my interval is between two same times, i.e. 12pm, maybe timezones?)

[antoinerousseau]

OK, all good now in both cultures, this fixes both bugs :)

@jquense can you take a look please?

It seems to also resolve the interrogations raised by @KerenChandran in #66 since the end date is included.

[jquense]

hi there thanks for the the contribution, I will take a closer look when I get a moment but a cursory glance has me I bit confused by what is going on and why the changes were made. In particular I don't think that math round is the right thing here... could you explain why you are using it? in general "because it made these dates correct" isn't great since there are A LOT of possibilities on a calendar and we need to not break other cases accidentally

[antoinerousseau]

Well, I see that the date diff calculation for slots (number of cells in a row) and span (number of cells the event is spanning on the current row) are not giving integers, e.g. 6.95833.. instead of 7, and I don't see why, since the diff method is called with day as the third argument. So it could be understandable for span if the start and end times are different, but what about slots? first and last always have the same time, right, since we are talking about entire days?

[jquense]

As I was diving into the PR here I ended up changing stuff and it ultimately was easier just to commit may changes. Thanks again for the contribution it helped!

[antoinerousseau]

Glad I could help! Thanks for the fix :)

You are also welcome to take some of the updates I made to the README, and also please add that missing & to the npm build script

And please publish on NPM :)

[jquense]

this isn't missing an & the single & is how you run parallel tasks on windows, I assume not so on Mac/*nix?

[antoinerousseau]

single & is also for parallel tasks on OSX/unix, but I don't think it's a good idea to run these tasks in parallel because if you wait for an exit code from the command, you'll get it before it completes, in addition to have loose output consistency (you don't know when exactly it finishes)...