vunit
vunit copied to clipboard
joaocunha
Add float classes
This update is creating 10 more classes for each existing class: x.1-x.9. giving user more precise classes.
Yeah sure. The difference between vwfs1 class and vwfs2 is way too much, 1vw is a lot (almost 20px on 1080 screens). If someone is in need for 1.5vw, he can now use it.
Actually, this change is based on a real case. Earlier today a friend of mine wanted to use your lib but he needed some font sizes between 1 and 2. I made this change originally for him, but then realized he's not the only one who can benefit from it, hence the pull request.
@Shebo I see what you mean. It's indeed not the first PR that tries to address this use case - which is a valid one.
The problem is that due the way the library works (it's event-less), nested for loops could be really expensive performance-wise.
I'm happy to accept a PR that addresses this issue, but I'm quite worried about the performance implications on a real website. Maybe we can benchmark the before and after?
I agree that this is not a perfect solution, it's quick n' dirty. Of course that a second nested loop will be expensive. Here's a quick test: http://jsperf.com/third-nested-loop/5
Looping 100xn times on every load/resize is a heavy operation as is, 10x100xn not helping in any form. I think there are no perfect solution without changing whole lib structre.
One possible course of action is to to have 1 nested loops that will run 10x100 times, like this:
for (var range = 0; range <= 100; range=sumFloat(range, 0.1))
It's not saving anything except another nested loop (test is showing no improvment: http://jsperf.com/third-nested-loop/6).
Another way is to make the loop asynchronous (like in jquery's/loadsh's/underscore's foreach), thus the browser loading time will not depend on these loops.