Remove support for the QuickDraw NPAPI drawing model

Remove support for the QuickDraw NPAPI drawing model

This model has been deprecated for a very long time, has very few plugins still relying on it, and is increasingly unsupported by other Mac browsers.

In addition to simplifying code, this ensures that plugin vendors with QD and CG implementations (which at least one has done as a short-term solution to get 64-bit Safari and Firefox support) don't leave Chromium on the QD path under the very mistaken belief that it's the faster path.

BUG=125913
TEST=Most Mac plugins should continue to work.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=143721

[stuartmorgan, 2012-06-21 12:17:22]

mark for review
jam for OWNERS

Mark, feel free to ping me if you want to discuss the current state of the
tradeoffs here, but I'm of the opinion that this is now the right choice.

[Mark Mentovai, 2012-06-21 14:47:36]

I’m nearly positive I agree, but could you just record the states of various
plug-ins as they exist now here on this code review for posterity?

[Avi (use Gerrit), 2012-06-21 14:51:55]

https://chromiumcodereview.appspot.com/10628004/diff/1/webkit/plugins/npapi/p...
File webkit/plugins/npapi/plugin_host.cc (right):

https://chromiumcodereview.appspot.com/10628004/diff/1/webkit/plugins/npapi/p...
webkit/plugins/npapi/plugin_host.cc:813: // We only support the Core Animation
model on 10.6 and higher
This comment isn't meaningful if our minimum is 10.6.

[stuartmorgan, 2012-06-21 15:12:02]

On 2012/06/21 14:51:55, Avi wrote:
> This comment isn't meaningful if our minimum is 10.6.

I know. There's a CL in flight that makes a bunch of changes in the CA
negotiation code so it should get fixed as a side-effect. If that doesn't
happen, I'll be doing a cleanup CL there. (The 10.5 support broke a while ago
anyway, so the comments in there were already wrong.)

[stuartmorgan, 2012-06-21 15:37:41]

On 2012/06/21 14:47:36, Mark Mentovai wrote:
> I’m nearly positive I agree, but could you just record the states of various
> plug-ins as they exist now here on this code review for posterity?

QD-only plugins I'm aware of in post-10.5 world:

- VLC (and by extension, Veetle): Has been broken for something like 9 months
because they don't have an info.plist; the've had a fix in their tree for about
that long, but never bothered to ship it.

- DivX: Has a CG mode that they decide to use based on UA instead of negotiated
models. I told them ~a year ago that this was a bad idea and was needlessly
killing performance in Chrome. I don't think they ever updated, but they could
at any time. I tried running it yesterday and it actually crashed deep in their
code during init, so unless something went wrong with my install it's already
broken in Chrome (and nobody has reported it).

- WebEx: Last I checked, never got an Info.plist despite pressure from Mozilla.
They were blocking at least some Chrome versions by UA on their site anyway.

- RealPlayer: Has never worked in Chrome. They have a CG-based shim for Safari,
but it's in a 64-bit version of the plugin instead of negotiating. If they
wanted to, they could trivially apply it to Chrome too.

- DjVu: Already broken due to not having an Info.plist. Nobody has ever reported
a bug about it.

- Move Networks: I don't know what the current state of this is, or if anyone
still uses it. At some point a year or two ago it stopped being directly
downloadable from them, and was just used by some ISP-specific content delivery
systems. If anyone is still using it, presumably they will update (or have
updated) for Safari and Firefox support anyway.

There's probably some more long-tail plugins, but I've never heard about them.

Also for posterity, the current state of browser support:
- Safari 64 silently doesn't load/display QD plugins, and as far as I can see
starting in 10.7 it doesn't load them even when started in 32-bit mode
- Firefox 64 puts a banner telling users they need to restart in 32-bit mode. My
understanding is that Firefox may well be fully removing QD support in roughly
the same timeframe as M22.

If someone really needs to use a QD plugin, they can always fire up Camino :D

[Mark Mentovai, 2012-06-21 15:53:38]

Thank you for the summary. I agree that it’s time to drop QD. LGTM.

[jam, 2012-06-21 16:20:20]

lgtm

[commit-bot: I haz the power, 2012-06-21 16:26:16]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/stuartmorgan@chromium.org/10628004/1

[commit-bot: I haz the power, 2012-06-21 16:26:22]

Presubmit check for 10628004-1 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for files in these directories:
    webkit

Presubmit checks took 3.6s to calculate.

Was the presubmit check useful? Please send feedback & hate mail to
maruel@chromium.org!

[stuartmorgan, 2012-06-21 16:29:37]

+darin for webkit/ OWNERS (for the gypi change)

[darin (slow to review), 2012-06-21 18:19:04]

LGTM for src/webkit/

[commit-bot: I haz the power, 2012-06-21 19:17:45]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/stuartmorgan@chromium.org/10628004/1

[commit-bot: I haz the power, 2012-06-21 19:55:43]

Try job failure for 10628004-1 on win for step "update".
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win&number...

Step "update" is always a major failure.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2012-06-21 20:25:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/stuartmorgan@chromium.org/10628004/1

[commit-bot: I haz the power, 2012-06-21 23:04:25]

Try job failure for 10628004-1 (previous was lost) (retry) (retry) on mac_rel
for step "browser_tests".
It's a second try, previously, step "browser_tests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[stuartmorgan, 2012-06-22 05:39:59]

Oh, right. The test plugin doesn't negotiate anything.

I'll have to make it a CG plugin and then come back to this :P

[stuartmorgan, 2012-06-22 08:02:01]

On 2012/06/22 05:39:59, stuartmorgan wrote:
> Oh, right. The test plugin doesn't negotiate anything.
> 
> I'll have to make it a CG plugin and then come back to this :P

That turned out to be a trivial change, so I just added it to this patch. Mark,
can you take a quick look?

(In answer to the obvious question "Why don't I also negotiate Cocoa while I'm
changing this": it looks like I'll need to set up an artificial run loop trigger
for that to work, since we lose the Carbon idle event timer and the run loop
appears not to be cycling in tests without it.)

[Mark Mentovai, 2012-06-22 17:54:16]

LGTM still.

[commit-bot: I haz the power, 2012-06-22 19:31:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/stuartmorgan@chromium.org/10628004/6002

[commit-bot: I haz the power, 2012-06-22 22:06:26]

Change committed as 143721