Removed generation step. Require directly from caniuse-db

[alexisvincent]

This passes the tests... I assume its equivalent to before

[alexisvincent]

Fixes #42

[MoOx]

Thank you so much for this. I will add some comment since I would like an option for "sort of" backward compat.

[MoOx]

Oh! I am seeing that caniuse now provides all data in a single json. That's cool and new (this file did not exist in the past).

I guess we should bump caniuse-db version to be sure data.json exist.

According to the change, does not look like a breaking change... Do you think we can release this as a patch? poke @Nyalab

[alexisvincent]

Yeah, didn't think it did.

[alexisvincent]

@MoOx Whether its a patch or not, do you think you could attempt to release a version of cssnext that uses this? Means I can close a bunch of issues all over the place :)

[MoOx]

@alexisvincent if it is released as a patch, this will be automatic for postcss-cssnext. If merged as a breaking change, I will have to bump, but yeah, I will do that if it's required.

As this merge/release could impact a lot of projects according to the number of download, I will wait to have some free times around (not a good idea a Friday afternoon) to cut a release, so if any major problems (or even minor) came up I can handle it quickly.

Again, thanks for taking time to handle this, I really appreciate the help!

[alexisvincent]

My pleasure. Thanks for the really tight feedback loop. Makes contributing easy. I just read through the entire codebase. Logic is sound. But found one features array access i didn't change (no test coverage). Everything is clean now. And is definitely not a breaking change.

[Nyalab]

can we test this on a small sized VM, to see if the perfs are not horrible? i'm afraid loading in memory all the db can make this step horribly long, but if it does not, we can release this as a patch @MoOx

[alexisvincent]

I can help with this if you'd like. Can run a docker container on my machine if that's ok with you?

[MoOx]

Lgtm

[MoOx]

@alexisvincent any update for the test in a slow env?

[alexisvincent]

Haven't had time yet. But will hopefully do it this afternoon. Unless one of you would like to.

[alexisvincent]

@MoOx I tested on google's f1-micro (1 shared vcpu, 600mb ram). Basically just time node list/index.js. Original had an average of 105ms, new one had an average of 112ms. So I think it's safe to say this is good to merge.

[MoOx]

@alexisvincent This is awesome. Will merge and release as a patch on monday morning! Thank you for your patience and involvement.

[alexisvincent]

👍

[Nyalab]

Thank you @alexisvincent & @MoOx

[MoOx]

Released as 1.5.2 \o/