487404 - The 'marker' attribute does not work

Status: RESOLVED DUPLICATE of bug 350100

(Core :: SVG, defect)

Description

[Jonathan Watt [:jwatt]]

The 'marker' attribute does not work.

Comment 1

[Jonathan Watt [:jwatt]]

Attachment: testcase

Comment 2

[Jonathan Watt [:jwatt]]

Attachment: patch

Comment 3

[Robert Longson [:longsonr]]

Comment on attachment 371660 [details] [diff] [review]
patch


Would be good to check in a reftest for this.

Comment 4

[Jonathan Watt [:jwatt]]

Yeah, the attached testcase is in reftest format. Should have included it in the diff, sorry.

Pushed http://hg.mozilla.org/mozilla-central/rev/aa5fd79f36b6

Comment 5

[Robert Longson [:longsonr]]

This is wrong. I'll back this out and correct the reftest. See bug 350100
and http://dev.w3.org/SVG/profiles/1.1F2/test/harness/htmlEmbed/painting-marker-04-f.html

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

The backout caused several testfailures

Comment 7

[Jonathan Watt [:jwatt]]

I'd rather get the spec changed than back this out. I think having a 'marker' attribute is good for authors.

Comment 8

[Robert Longson [:longsonr]]

I've fixed the testfailures.(In reply to comment #6)
> The backout caused several testfailures

Opera ignores the marker property.

I think we should leave it backed out while you raise it with w3c. We can always flip it yet again.

Comment 10

[Robert Longson [:longsonr]]

pushed http://hg.mozilla.org/mozilla-central/rev/cab568cf371d and
http://hg.mozilla.org/mozilla-central/rev/32b6af68ddb2

Comment 11

[Jonathan Watt [:jwatt]]

I'm also not very keen of having changes like this pushed without discussion or review. Although backing out aligns with our historic stance of following the spec to the letter no matter how silly it is, we've been taking a more pragmatic view in recent years where it makes sense.

And why mark this as a duplicate of a bug closed INVALID? That just pushes the issue off the radar.

Comment 12

[Robert Longson [:longsonr]]

Traditionally we haven't had reviews for backouts but that doesn't mean I'm not willing to change.

I'm concerned we're the odd one out here since at the very least Opera and Batik follow the spec, that's yet another gotcha for authors currently if they develop against us.

I think you should reopen the bug 350100 if you're taking this issue to the w3c.

Comment 13

[Robert Longson [:longsonr]]

In fact I just did reopened bug 350100.

Comment 14

[Jonathan Watt [:jwatt]]

I take your point about author gotchas, but I don't think we're dealing with the normal backout case due to unexpected problems discovered right after checkin here.

In this case the checkin and bug closure occurred over 14 months ago, and in fact the patch has shipped in Firefox 3.6 and other products.

I'd also note that some of the demos accompanying the talk that I gave at SVG Open 2009 are broken by this backout.

Comment 15

[Jonathan Watt [:jwatt]]

I'm torn because I _do_ want to try and avoid bug filing and review overhead where it's pointless and just wastes dev time, but for something like this I thing a quick check with one of the other module peers would have been nice.

Anyway, I guess it's just one of the expected bumps along the road to more efficient working. Onwards to bug 350100.