Page MenuHomePhabricator

Check Tangram browser support against WMF support matrix
Closed, ResolvedPublic

Description

Tangram JS requires a reasonably recent browser for full support. WMF has full support for reasonably recent browsers. The two definitions for reasonably recent might differ and should be checked.

In practice, Tangram JS should also run in any browser with WebGL support.

Event Timeline

https://www.mediawiki.org/wiki/Compatibility
https://github.com/tangrams/tangram#browser-support

Tangram structures its support by OS and browser, while for desktop MW support is only by OS. The following do not have official support in Tangram, but are Modern (Grade A) for MediaWiki core (and extensions that decide to follow it)

  • Opera
  • IE 10
  • Safari 5.1+ for Windows (not available from Apple anymore)
  • OSes other than Windows or Mac OS

I'm not too worried about Opera, since it uses he same rendering engine as others, or Safari 5.1+ for Windows, which is near-impossible to get. Other OSes is not likely to be a problem, even if they're not officially supported by Tangram. I'm also somewhat skeptical about our claims to actively test against combinations like Opera on FreeBSD.

The big difference is that Tangram will not run on IE10, which does not support WebGL. Do we need to worry about this?

Tangram has official support for Edge on Windows, which is something beyond the MW Core support.

In general, Tangram should run on anything with WebGL, even if they don't test it.

I'd expect Mobile Web support to be similar, but can't tell, since https://www.mediawiki.org/wiki/Compatibility is an inconsistent confusing mess for mobile. It references Grade B, which isn't defined, has conflicting requirements with different mentions of the same browser, and doesn't clearly differentiate between browser and OS.

IE10 is the only real issue I see, and I recall hearing something about dropping support for it in the future. As the matrix only applies to MW core, I don't see any issues with any other minor deviations from the matrix, when the overall intent is the same.

tl;dr: probably okay, except IE10.

Gehel added a subscriber: jmatazzoni.

Moving to "done" as the analysis is completed and looks good.

@Fjalapeno / @jmatazzoni : I'll let you close that task unless you want more precision in the analysis (unlikely).