[webkit-gtk] Removing support for forever-experimental features

Michael Catanzaro mcatanzaro at igalia.com
Sat Mar 18 08:35:59 PDT 2017


We have several forever-experimental features that have had zero or
almost zero progress in several years:

 * Accelerated 2D canvas
 * GStreamerGL
 * WebCrypto

It's really not acceptable to keep prototypes of failed features in
trunk when there is no foreseeable path to ever finishing them. (These
are in contrast to actively-developed experimental features, like
WebRTC, which are not problematic.) I would like to remove all the code
for these features (except for the cross-platform WebCrypto stuff that
Apple needs, of course). If someone wants to commit to actively
developing the features in the future, then and only then should we
bring it back. In the meantime, it's just getting in the way and
causing extra gardening effort.

 * Accelerated 2D canvas used to be enabled only for bots and tests,
but at the last hackfest we decided to disable it even there. So nobody
is using it or working on it. It's dead code. Why keep it around? If we
want it in the future, then we can bring it back.
 * GStreamerGL and WebCrypto are enabled only for bots and tests, but
this is pointless as nobody is working on new test failures for
multimedia layout tests or for WebCrypto. Anyone who wants to work on
these in the future is going to have to go through all the tests
individually anyway. In the meantime, it's just making it harder to
garden. *Especially* WebCrypto, since our WebCrypto support just keeps
getting more and more broken as Apple implements it properly on their

If multimedia folks think they can get GStreamerGL stable soon, then
maybe we can keep it, but it's time to get serious about it. But I'm
pretty sure nobody has any interest in developing accelerated 2D canvas
or WebCrypto right now.


More information about the webkit-gtk mailing list