130605 - Java applet doesn't load on www.rav.uplink.pl

Status: VERIFIED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Gabriel Baines]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; WinNT4.0; en-US; rv:0.9.9) Gecko/20020311
BuildID:    2002031104

In the leftmost frame on this page: http://www.rav.uplink.pl/ there is a Java
applet (menu). 
It works perfectly with: Mozilla 0.9.8 (Build ID: 2002020406) + JDK 1.4
It doesn't work with: Mozilla 0.9.9 (Build ID: 2002031104) + JDK 1.4
Both builds tested on the same box, with same JDK.

On both builds other pages with Java seem to work OK, it is a problem with this
particular page. However, because it seems like regression (works on 0.9.8) I
file it as a browser bug.

If this is a problem with the page applet tag, please change product -> Tech
Evangelism


Reproducible: Always
Steps to Reproduce:
1.Go to http://www.rav.uplink.pl/
.

Actual Results:  Java does not start resulting in no menu in the left frame.

Expected Results:  Java starts, applet gets loaded and initialized, menu apperas
in the left frame.

Comment 1

[Igor Nekrestyanov]

Applet in left frame does not start for me too 
(Linux with jre1.3)
Confirming.

Looks like problem with parsing (left frame has incorrect html)
NS 4.x shows it but i am not sure 0.9.8 did this 
(html probably was changed). Anyway, problem is worth looking to it.

Comment 2

[Gabriel Baines]

I'm absolutely sure it works with 0.9.8 - I've just doublechecked it on builds
2002020406 and 2002020409 on two different boxes (both WinNT). 

Comment 3

[Denis Antrushin]

This is not parser problem. 
Taking...

Comment 4

[Denis Antrushin]

This regression is a side effect of darin's fix for bug 124042.
Before we didn't use NS_NewURI, which it turn calls nsIIOService::NewURI, 
which at some point call nsStandardURL::Init().
And in Init() there are lines:
  if (spec.IsEmpty()) {
       Clear();
       return NS_OK;
  }
In our case, spec is value of codebase attribute and when it's empty
(codebase=""), we get empty url in nsObjectFrame.cpp line 1066 and
then call InstantiatePlugin with null fullURL, which, finally, causes
'No Plugin Downloaded Plugin' to popup.
I don't think that nsStandardUTL::Init() is wrong, so fix would be to check
codebase before calling MakeAbsoluteURL

Comment 5

[Denis Antrushin]

Attachment: suggested fix, v1

don't call MakeAbsoluteURL for empty codeBase

Comment 6

[Denis Antrushin]

I think this affects other plugins as well. Moving to Plugins
Peter, could you review this?

Comment 7

[Peter Lubczynski]

Comment on attachment 74316 [details] [diff] [review]
suggested fix, v1

r=peterl

Comment 8

[Denis Antrushin]

While improving patch I found few lines which totally confused me:
1017    if (NS_CONTENT_ATTR_HAS_VALUE == mContent->GetAttr(kNameSpaceID_HTML,              nsHTMLAtoms::codebase, codeBase)) {
1018    nsCOMPtr<nsIURI> codeBaseURL;
1019    rv = MakeAbsoluteURL(getter_AddRefs(fullURL), codeBase, baseURL);
1020    if (NS_SUCCEEDED(rv)) {
1021        baseURL = codeBaseURL;
1022    }
We're not initializing codeBaseURL in line 1018, so what the reason
of assignment in line 1021? Shouldn't line 1019 read
rv = MakeAbsoluteURL(getter_AddRefs(codeBaseURL), codeBase, baseURL);
???

Comment 9

[Denis Antrushin]

Attachment: fix, v2

Better fix. 
The same manipulations were done in three places before.
Now baseURL handled in single place. Also same trick done
with URL construction from 'src' and 'data' attributes
(there was two identical pieces of code under different |if|s.)

Comment 10

[Denis Antrushin]

Peter, could you review latest patch, please?

Comment 11

[Peter Lubczynski]

Comment on attachment 74714 [details] [diff] [review]
fix, v2

r=peterl

Comment 12

[Patrick C. Beard]

Comment on attachment 74714 [details] [diff] [review]
fix, v2

sr=beard

Comment 13

[Asa Dotzler [:asa]]

Comment on attachment 74714 [details] [diff] [review]
fix, v2

a=asa (on behalf of drivers) for checkin to the 1.0 trunk

Comment 14

[Peter Lubczynski]

patch in trunk, marking FIXED.

Comment 15

[Denis Antrushin]

*** Bug 133395 has been marked as a duplicate of this bug. ***

Comment 16

[Patty Mac]

Verified on windows 98 commercial trunk (2002-04-15-10-TRUNK)