Convert operator and type names to lowercase with underscores.

[sunfishcode]

Yet another attempt to resolve #83.

[lukewagner]

Also me. On pure personal preference, lower_with_underscores is my favorite naming convention. But it's also good by avoiding the uncanny valley of being 1 capitalization away from two popular naming schemes.

[titzer]

Runs faster too, because C.

On Tue, Jun 9, 2015 at 4:55 PM, Luke Wagner [email protected]
wrote:

Also me. On pure preferences, lower_with_underscores is my favorite naming
convention. But it's also good by avoiding the uncanny valley of being 1
capitalization away from two popular naming schemes.

—
Reply to this email directly or view it on GitHub
WebAssembly/spec#122 (comment).

[lukewagner]

Another reason: there is necessarily going to be a mapping from WebIDL signatures into wasm signatures and this mapping will (I hope!) take Uint32Array not to some wasm view type, but to two parameters uint32* base, uint32 length (thereby avoiding temporary garbage). Thus, Uint32Array and the rest of the view types are necessarily separate kinds of things from anything in wasm.

[kripken]

Typed arrays will still be passable to wasm, though, as mentioned here? I don't think we can avoid Uint32Arrays etc. being a name that comes up in wasm.

Overall, this seems strictly worse than to follow existing web conventions, but if I am the only person that thinks that way, this pull seems to me like the least bad of all alternatives.

[lukewagner]

@kripken Yes, that would be a separate way for wasm to operate on a real JS Uint32Array.

[sunfishcode]

Ok, I closed #83 and #93 so that we can focus on this one approach. Do people like this? Dislike it?

Some of the names here, in particular load_heap_with_offset, int32_from_float32_bits, but possibly others too, may want further renaming, but the main goal here is to get consensus on the overall style first.

[jfbastien]

I like this.

[MikeHolman]

I agree with @kripken that I would prefer we follow existing web conventions (e.g. by using Uint32 style), but on the other hand I don't think it's a big deal.

[jfbastien]

Is there an outcome on this?

[kripken]

@pizlonator is the only one I think I remember participating in the other discussions on this, that has not weighed in here with an ok. @pizlonator, thoughts?

[sunfishcode]

Now updated to use dots instead of underscores when it looks natural to do so, so operators now look like int32.add and float32.mul, since reactions to this among people I've shown it to so far were all positive.

[jfbastien]

As discussed on IRC: merging now so we can move forward and merge dependent PRs (#135, further refactoring); we can revisit naming later.