Bug 17494 – Adobe SVG test case rejected

• Status: RESOLVED
• Resolution: WONTFIX
• Component: SVG

[cnep5ll3]

Launch WebKit 30468
Error messages in space for rendering.

This is some sort of fancy installation test by Adobe.
Safari 3.0.4 (only) does this well.

The following all fail:
Opera 9.24
Firefox 2.0.0.12
Camino 2007022813 (1.0.4)

[Alexey Proskuryakov]

I'm surprised that this test works for you with Safari 3.0.4 - it doesn't for me. Have you by any chance installed Adobe SVG viewer and changed Safari to run under Rosetta?

The root cause here is that the test is sent as text/plain, while it is actually encoded as gzip. An underlying system framework mistakenly sniffs this as image/svg+xml for us. I have filed <rdar://problem/5765467> to track this.

According to bug 5246 comment 10, not making a workaround for this server bug was an intentional decision by browser makers. So, I'm closing this as WONTFIX (as the test will probably still be rejected after the bug I filed with Apple is fixed, and we want to match other browsers).

[Norman Hardy]

About one year ago I installed an SVG viewer from Adobe.
Just now I spent 15 minutes trying to learn, with Safari Help, how to determine which plug-ins were active or available.
I failed. I think Help is out of date. I will file a Help bug report if you can recommend a place to do so.
I do not recall taking any steps the run Safari under Rossetta; Safari seems very responsive.

[Alexey Proskuryakov]

Do you see Adobe SVG in Safari 3.0.4 in Help->Installed Plug-ins?

AFAIK, this plug-in is PowerPC-only, so one needs to run Safari in Rosetta in order to use it. You can try Get Info on Safari.app to see is Rosetta checkbox is enabled.

> I will file a Help bug report if you can recommend a place to do so.

<http://bugreport.apple.com> is the place for all Apple-specific bugs, so help issues should go there. One free ADC Online membership to login: <http://developer.apple.com/products/>.

[Norman Hardy]

Created attachment 19396 [details]
A Grab of the Info window for Safari in the applications folder

Note version, and this it is not under Rosetta.

[Norman Hardy]

Created attachment 19397 [details]
rendering of SVG test case by Safari 3.0.4

This is a Grab of Safari correctly rendering the Adobe installation test case.
The image was indeed evolving even as Grab captured the image.

[Norman Hardy]

Perhaps Safari, seeing a PPC Plugin, opts for Rosetta even without the check box in the app info window.
There was a pause sufficient for the Rosetta transmogrifications.
If this were so, WebKit might need a similar function.

[Alexey Proskuryakov]

(In reply to comment #6)
> Perhaps Safari, seeing a PPC Plugin, opts for Rosetta even without the check
> box in the app info window.

I don't think so - plug-ins are looked at much later.

Count me thoroughly confused. Could you please attach the list of plug-ins in Safari 3.0.4 (Help->Installed Plug-Ins, File->Save As)?

[Norman Hardy]

Created attachment 19402 [details]
list of Plug-ins for Safari 3.0.4

Here are the Plug-ins. Adobe's NPSVG3Carbon is included!

[Alexey Proskuryakov]

My bad - I didn't realize that Save As didn't work - I'm seeing my own list of plug-ins :)

Could you please copy/paste the list?

[Norman Hardy]

Created attachment 19403 [details]
Better list of PlugIns.

I should have noticed. Sorry.

[Alexey Proskuryakov]

So, so the list shows the usual Adobe SVG 3.0 plug-in, which was released many years before Apple switched to Intel processors. The only way Safari could use it on Intel is if it were running in Rosetta.

Could you please check the Safari application type in /Applications/Utilities/Activity Monitor.app? It's the rightmost column in process list.

[Norman Hardy]

Activity Monitor reports Safari running in PPC mode despite lack of Rosetta check in Info box.
Activity Monitor reports WebKit running in Intel mode despite Rosetta check in Info box.
(just a test, not normally checked!)
This reduces the mystery to a lesser mystery: How does Finder (launchd?) decide whether to use 
Rosetta?
This is presumably not a WebKit issue.

I have learned a lot. Thanks!