425752 - Image grid not rendering

Status: RESOLVED WORKSFORME

(Web Compatibility :: Site Reports, defect)

Description

[Michal Urban]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9b5) Gecko/2008032619 Firefox/3.0b5
Build Identifier: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9b5) Gecko/2008032619 Firefox/3.0b5

Please see details from FX3 Beta 5 test day in "Additional Information".

This page is generated by Apple's iWeb application. It was affected earlier by the FX 2.0.0.10's canvas regression:

http://www.mozillazine.org/talkback.html?article=22741

Reproducible: Always

Steps to Reproduce:
1. load the page, image grid will not render
2.
3.
Actual Results:  
Only the top image will show.

Expected Results:  
A grid of 3x5 images expected under top image. Verify the same page in FX2 or Opera.

Transcript of the related chat at irc.mozilla.org#testday, March 28 2008

NikonMike: anybody knows why this page won't work in FX3?
NikonMike: http://oakvillecameraclub.com/Competition.html
NikonMike: generated by Apple iWeb, uses canvas (I think)
NikonMike: there's supposed to be a 3x5 grid of images...
tchung: nikonmike: hmm
tchung: does it work on fx2?
NikonMike: yes, it does!
NikonMike: also in Opera, Safari
tchung: looks like an error is thrown
tchung: Error: widgetDiv.getElementsByClassName("paginated_only").invoke is not a function
tchung: Source File: http://oakvillecameraclub.com/Scripts/Widgets/HeaderControls/HeaderControls.js
tchung: Line: 100
tchung: go ahead and bug it
NikonMike: I think I already reported it as "broken web-site"
NikonMike: should I open a new bug in Bugzilla?
tchung: i dont think its website though
tchung: it could be something we need to fix on the platform
tchung: mw22: any ideas where a bug like this would go>?
Tomcat: tech-evanglism ?
mw22: yeah, that seems like a tech evangelism iss ue to me
mw22: although.. don't we support getElementsByClassName now?
mw22: ah, I see, the invoke after it, is what is causing trouble
mw22: it seems Protoype is using .invoke
mw22: I guesss now that Firefox 3 knows about getElementsByClassName, the Prototype version of it isn't used anymore, but then it trips over the .invoke function
tchung: ah i see
tchung: so it sounds like it should be tech-evangelism then
tchung: since web authors need to be informed of this class
mw22: yeah, it still seems like a tech-evangelism issue to me
NikonMike: isn't it easier for you to report it (tchung, mw22) now that you analyzed it?
NikonMike: not that I don't want to help
mw22: NikonMike, you can report it, just CC me on the bug
NikonMike: ok then
mw22: NikonMike, my mail address is martijn.martijn(at)gmail.com
mw22: Just try to add all the relevant info to the bug
NikonMike: ok, I'll put this info there
tchung: nikonmike, yeah no worries, we'll comment in the bug.   my cc is [e-mail]
tchung: thanks for filing!

Comment 1

[Tony Chung [:tchung]]

bug confirmed.

Error: widgetDiv.getElementsByClassName("paginated_only").invoke is not a function
Source File: http://oakvillecameraclub.com/Scripts/Widgets/HeaderControls/HeaderControls.js
Line: 100

Comment 2

[Chris Lawson (gone)]

Am I correct in understanding this is something that Apple is going to need to change in the iWeb app itself, lest all iWeb-created sites continue to be broken in future Gecko browsers?

If that's the case, I think an elevation of severity above "normal" would be in order here.

cl

Comment 3

[Michal Urban]

Thanks for your comments, Chris!

Since this code works in all current browsers (Firefox 2/Flock, Safari/WebKit/Shiira, Opera, MSIE 6 and 7) it's difficult to accept that it's just a "coincidence" and Firefox 3 is "right" by not supporting it.

Hopefully, not all iWeb sites would be broken, depending on what templates/controls are actually being used.

Comment 4

[Boris Zbarsky [:bzbarsky]]

It's odd that Safari doesn't have this problem, since it implements getElementsByClassName.  Can someone minimize a testcase that works in Safari here but fails in Gecko?

Comment 5

[John Resig]

Isn't this the same as bug 390411? They're using an old version of Prototype - and we don't allow for overriding getElementsByClassName.

Comment 6

[Boris Zbarsky [:bzbarsky]]

Ah, they set it on the prototype.  That might explain why things work in Safari, yeah.

Comment 7

[Boris Zbarsky [:bzbarsky]]

For what it's worth, what confuse me here was bug 412150.  Looks like Prototype 1.6 does the check before assigning its own thing, but Prototype 1.5 (and presumably this page?) doesn't.  The case when it does the check would fail in Safari too....

Comment 8

[David Glover]

This appears to be a duplicate of bug 405903.

Comment 10

[Boris Zbarsky [:bzbarsky]]

Uh... no.  This is an _evangelism_ bug.  Which means it needs to be fixed by the site.  Which means that different sites == different bugs.

Comment 11

[Patrick]

(In reply to comment #10)
> Uh... no.  This is an _evangelism_ bug.  Which means it needs to be fixed by
> the site.  Which means that different sites == different bugs.


I understand that, but it would be a bit unpractical to have on bug for each site created with iWeb 2.x. Also, some of these bugs contain a good description of the problem, for instance bug 437312 comment 0 filed by someone from Apple.com. Isn't it better to in some way collect these bugs which all have their root in the same cause?

Comment 12

[Boris Zbarsky [:bzbarsky]]

The goal is to have all these sites get fixed.  That means one bug per site...  If you collect them all and, say, get iWeb fixed, you still have to contact all the sites and get them to upgrade to the fixed version.

So perhaps marking dependencies makes sense.  Duplication of evangelism bugs does not.

Comment 13

[Bastiaan van de Werk]

I seem to have the same problem on this site:

http://www.vandewerk.nl/vandewerk/Caspar_2.html

Comment 14

[Chris Lawson (gone)]

Bastiaan: please file a *new* bug on that site; see comments 10 and 12.

In the meantime, please let the webmaster know that it's broken and that there's a fix available in the latest iWeb from Apple. He just needs to go through and re-save all the pages and re-publish them.

Comment 15

[Karl Dubost💡 :karlcow]

The site for which this bug had been opened doesn't exhibit the issue anymore.