Future of X- Prefixes

[mitsuhiko]

I have no strong opinions here but I came across this again. HTTP no longer recommends x- prefixes, the future of x_ prefixes in this specification should maybe be reconsidered.

At least I would argue that we should not recommend vendor prefixes. If we think for instance that ignoreList is a useful thing, it's odd that all tools are supposed to refer to this as x_google_ignoreList.

[littledan]

We use some vendor prefixes at Bloomberg. I'm not really sure if vendor prefixes are a bad thing here.

[bmeurer]

My 2 cents here is that vendor prefixes are a quite useful way to experiment with new ideas. Let's take the case of x_google_ignoreList: The vendor prefix allowed us to experiment with this and get this in front of real users to understand whether we are really solving the right problem here, and we can now bring that understanding to the standards body to agree on how to properly standardize it (which could be as simple as removing the x_google_ from the beginning and having tools support both x_google_ignoreList and ignoreList for a few releases).

[mitsuhiko]

I guess I'm devil's advocate here but I think the fundamental findings of RFC 6648 which recommends against the X- prefix for new experimental HTTP headers apply to source maps too (See appendix B).

[jkup]

I think a lot of people like the freedom for vendors to experiment with x-prefixes but, as Armin says, there have been a ton of cases where people have "experimented" with an x-prefix and then the thing gets standardized and then we have to go through and get rid of the x-prefix. Like is the case now with the x_google_ignoreList which we'd like to move and rename (since Firefox uses it too).

Maybe the best way forward is to work together on moving the x_google_ignoreList into a more official name, see how we like that process and then circle back on our recommendations on x-prefixes in general?

[jkup]

Since Google and Mozilla seem happy with this, can we add an official ignoreList to the spec? @mitsuhiko brought up a good point about what to do if both show up. Let's talk about this further.

For x-prefixes in general, this seems like it merits more discussion.

[EricSL]

For the x_google_ignoreList change, I made a pull request: tc39/source-map-spec#19

[littledan]

Will be fixed by tc39/source-map-spec#24