220776 - [FIX]Frameset in jar file failed to be shown

Status: RESOLVED FIXED

(SeaMonkey :: General, defect, P1)

Description

[Steve Liu]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925

If all jdk api help is in a jar file.
open it with jar protocol and with "frameset" mode.
you can see all the frame shown blank 

Reproducible: Always

Steps to Reproduce:
1.Locate:   jar:file:///D:/JBuilder9/doc//jdk_docs.jar!/java/api/index.html
2.
3.

Comment 1

[Boris Zbarsky [:bzbarsky]]

Please attach a testcase to this bug.

Comment 2

[Steve Liu]

I am not sure if here is the right place to post a testcase... 
anyway:

1. Find a jar file consists of some JAVA api, SUN's JDK api is an example easy
to be found. Lets suppose it is "D:\JBuilder9\doc\jdk_docs.jar"

2. Now, visit  
"jar:file:///D:/JBuilder9/doc//jdk_docs.jar!/java/api/index.html" in Mozilla 1.5 RC2

3. You can see there are 3 frames shown, but all of them shown empty.




This bug does not exist in Mozilla 1.4 I used before. Mozilla 1.4 will shown 3
frames with right contents in them.

Comment 3

[Steve Liu]

Easier steps to  reproduce this bug:

1. In "C:\temp" create 3 files
   c:\temp\index.html
   c:\temp\left.html
   c:\temp\right.html
2. edit "c:\temp\index.html" to:
   <pre>
      <frameset cols="50%,50%">
	<frame src="left.html">
	<frame src="right.html">
    </frameset
   </pre>
3. edit "c:\temp\left.html" to:
    <pre>
      <h1> This is LEFT</h1>
    </pre>
4. edit "c:\temp\right.html" to:
    <pre>
       <h1> This is RIGHT</h1>
    </pre>
5. Open "C:\temp\index.html" in Mozilla 1.5 RC2, you will see 2 frames  
showing "This is LEFT" and "This is RIGHT".

6. Now, jar them together:
    <pre>
     cd c:\temp
     jar cvf test.jar *.html
    </pre>

7. Open index.html in test.jar within Mozilla 1.5 RC2
    jar:file:///C:/temp/test.jar!/index.html
   You will see there are 2 frames showing nothing.


Again, this bug does not exist in Mozilla 1.4

Comment 4

[Steve Liu]

Very sorry, I didn't know it was unneccessary to add "<pre>..</pre>"
to html sections in my post.  it is the first bug I post here.

So, the right easier testcase should be:


1. In "C:\temp" create 3 files
   c:\temp\index.html
   c:\temp\left.html
   c:\temp\right.html
2. edit "c:\temp\index.html" to:
      <frameset cols="50%,50%">
	<frame src="left.html">
	<frame src="right.html">
    </frameset
3. edit "c:\temp\left.html" to:
      <h1> This is LEFT</h1>
4. edit "c:\temp\right.html" to:
       <h1> This is RIGHT</h1>
5. Open "C:\temp\index.html" in Mozilla 1.5 RC2, you will see 2 frames  
showing "This is LEFT" and "This is RIGHT".

6. Now, jar them together:
     cd c:\temp
     jar cvf test.jar *.html

7. Open index.html in test.jar within Mozilla 1.5 RC2
    jar:file:///C:/temp/test.jar!/index.html
   You will see there are 2 frames showing nothing.


Again, this bug does not exist in Mozilla 1.4

Comment 5

[Boris Zbarsky [:bzbarsky]]

Since you seem to have a jar file showing the problem, could you just attach it
to this bug using the "Create a New Attachment" link.

Comment 6

[Steve Liu]

Attachment: a jar contains 3 html files.

Comment 7

[Steve Liu]

Download the attachment and save it as "C:\temp\test.jar"

Visit 
jar:file:///C:/temp/test.jar!/index.html

in both Mozilla 1.5RC2 and Mozilla 1.4 to see the difference.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Attachment: Patch

This is caused by the change to the frame-detection algorithm that assumes that
all URIs will succeed when you call GetPath() on them.	Jar URIs happened to
fail. This patch fixes them to return a useful path for GetPath().

Comment 9

[Boris Zbarsky [:bzbarsky]]

Taking.  This is a pretty simple change that we may want for 1.5 to unbreak
framesets in jar files.  Last I checked, IBM & co. were pretty keen on things
like that....

Comment 10

[yonatan goraly]

I believe that it is related to this bug - a JAR url is not loaded into a
frameset as well. Since this is a workaround to execute signed code from a
remote dynamic (jsp) application, this bug will totally break such an application.

Comment 11

[Boris Zbarsky [:bzbarsky]]

Fix checked in.

Comment 12

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 132481 [details] [diff] [review]
Patch

Could this please be approved for 1.5?	This just makes jar: uris not throw an
exception when you try to get the path from them.  The path is used to prevent
runaway frameset recursion, and the error return prevents frames which point to
jar: uris from working....

Very safe patch that only affects people who call GetPath() on a jar: uri (and
only adversely affects those who expect such a call to fail).

Comment 13

[Asa Dotzler [:asa]]

Comment on attachment 132481 [details] [diff] [review]
Patch

1.5 shipped. removing obsolete request.

Comment 14

[Matthias Versen [:Matti]]

*** Bug 224056 has been marked as a duplicate of this bug. ***

Comment 15

[Olivier Cahagne]

*** Bug 245438 has been marked as a duplicate of this bug. ***