Opera v4.0 Mangling Linked URLs
after...@global-tetrahedron.com
This link will take you to a particular thread on one of these message boards which has multiple pages. These multiple pages are represented at the top of the page, with links to the other pages, like so:
Pages in this thread 1 | 2 | 3 | 4 | 5 | 6 | 7 | (show all)
If you are on the first page of this 7-page document, you can click on 2 to take you to the second page, 5 to take you to the fifth page, etc. The "Show all" link will display the entire thread as one large page.
Under Opera, when I try to click on these links, Opera mangles it. For example, if I were to click on the "show all" link I *should* be taken to the page:
This is the link copied verbatim from the source code of the page.
In actual fact, when I click on the "show all" link, I am confronted with the same page I was looking at. And if I right-click on the "show all" link and copy the link address, it gives me this URL instead:
http//boards.stratics.com/cgi-bin/boards/eq/showflat.pl?Cat=7&Board=eqwestwood&Number=262469&page=0&view=collapsed&sb=5 =all&vc=1
Note the odd "box" symbol between "sb=5" and "=all" there towards the end. For some reason, Opera is parsing the "&part" of the actual URL as some sort of control character, and it does so consistently.
Is this merely a case of bad HTML programming? Is "&part" a reserved part of HTML? If so, shouldn't Opera parse it as the reserved character only if it has a semi-colon as a closing delimiter?
Any insight you can provide into this issue would be most appreciated. Thanks in advance.
AB
Dines Justesen
different sites. The problem is that is the links uses &s instead of &s
O will replace the &s with a charentity. For the &s to be valid char
entities a semicolon would be needed, so I think the best way to handle that
would be not to replace them unless the ; is there. OS has decided to
replace that charentities even when no ; is there, and when I suggested it
was changed Sue pretyt much said that this was not going to happen. The only
"solution" to this problem is therefore to use another browser for viewing
all the sites which has this problem.
Dines
--
-----------------------------------------------------------------------
Dines Justesen
"There are two major products that come out of Berkeley: LSD and UNIX.
We don't believe this to be a coincidence." -- Jeremy S. Anderson
Email: di...@aub.dk
WWW: www.aub.dk/~dines
-----------------------------------------------------------------------
<after...@global-tetrahedron.com> wrote in message
news:1103_964786350@he...
Sue Sims
wrote:
>This has been discussed before as it causes lots of probems with many
>different sites. The problem is that is the links uses &s instead of &s
>O will replace the &s with a charentity.
Sorry, but I'm having difficulty parsing this: O will replace the &s
with a charentity. Do you mean that Opera will interpret the '&' as
the beginning of an entity reference? If so, then we are back to a
discussion of whether '&' is a reserved character, with special
meaning, and in which context that meaning should apply.
>...OS has decided to
>replace that charentities even when no ; is there, and when I suggested it
>was changed Sue pretyt much said that this was not going to happen.
I remember an interesting discussion of whether unescaped ampersands
were legal in any portion of the path, but I think you inferred
something unintended in my portion of the debate/discussion. I do not
have a voice in making policy decisions within Opera Software, and
have tried my best to refrain from saying something-or-other would (or
would not) happen, except when citing Operafolk who *do* have a voice
in the policy making process.
In the current build I'm using/testing,
<url:http://boards.stratics.com/cgi-bin/boards/eq/showflat.pl?Cat=7&Board=eqwestwood&Number=351561&page=0&view=collapsed&sb=5>
leads directly to the message board.
>The only
>"solution" to this problem is therefore to use another browser for viewing
>all the sites which has this problem.
Alternatively, I'd suggest downloading Opera 4.02 when it becomes
available.
Sue
Ken Knox (N1JRO)
[snip]
>
>In the current build I'm using/testing,
><url:http://boards.stratics.com/cgi-bin/boards/eq/showflat.pl?Cat=7&Board=eqwestwood&Number=351561&page=0&view=collapsed&sb=5>
>leads directly to the message board.
>
It does here too with 4.01.
[snip]
Ken
- Opera 4.01.0.754, Win32 - Identifying as Opera unless noted!
- WebWasher 2.1.1
- AMD K6-266 - W95b - Mirage Illusionist 3D Video (32M) - 192M SDRAM
- Sun Java JRE 1.3, Real G2, Adobe Acrobat 4.0, & QuickTime plugins.
- Running thru local 10 Base-T server network(P133, USR 56kModem)
Dines Justesen
> discussion of whether '&' is a reserved character, with special
> meaning, and in which context that meaning should apply.
This problem has been disccused several times before. &s can have special
meaning, but is also allowed to be used as a normal char.
> I remember an interesting discussion of whether unescaped ampersands
> were legal in any portion of the path, but I think you inferred
> something unintended in my portion of the debate/discussion. I do not
> have a voice in making policy decisions within Opera Software, and
> have tried my best to refrain from saying something-or-other would (or
> would not) happen, except when citing Operafolk who *do* have a voice
> in the policy making process.
Your statement did indicate to me that it was unlikely to be fixed. If it is
going to be fixed that is great as it causes lots of problems as it is now.
> In the current build I'm using/testing,
>
<url:http://boards.stratics.com/cgi-bin/boards/eq/showflat.pl?Cat=7&Board=eq
westwood&Number=351561&page=0&view=collapsed&sb=5>
> leads directly to the message board.
That link works in 4.01 too. The interesting part is whether the links on
the page mentioned in the original post works or not.
> Alternatively, I'd suggest downloading Opera 4.02 when it becomes
> available.
Well until 4.02 becomes available that doesn't really help most people.
Dines
Afterburner
>In the current build I'm using/testing,
><url:http://boards.stratics.com/cgi-bin/boards/eq/showflat.pl?Cat=7&Board=eqwestwood&Number=351561&page=0&view=collapsed&sb=5>
>leads directly to the message board.
It should, since that URL isn't broken. =)
However, go to that page and click on the "show all" link. What SHOULD
happen (in an ideal world) is that when you click this link, you are taken
to a page that contains all 350+ individual messages which make up that
entire thread.
What ACTUALLY happens is that the page essentially reloads an identical
copy of itself. You see activity as though the page were loading a new
page. But when it finishes, you're still staring at Page 1 of <X>, with
only the first 50 messages from the 350+ message thread.
As near as I can figure, this occurs because the "show all" URL contains
the string "&part".
The problem seems to arise in Opera's conversion of the source code of the
*first* URL above.
Try this as an experiment:
1) Go to the URL at the top of this message.
2) View the source code of the page. Find the source code for the "show
all" link. Copy the URL that the "show all" link should link to. Paste
this somewhere in a text editor.
3) Close the source code.
4) Move your mouse over the "show all" link.
5) Right-click and select "Copy link location" from the pop-up menu. Paste
this link location into the same text editor as in step 2.
6) Compare.
7) Note that they are not identical URLs. The URL that you retrieve from
actually looking at the source code is:
URL#1:
The URL you retrieve from right-clicking on the link and selecting "copy
link location" is:
URL#2:
Hopefully this makes sense. The URL that Opera returns when you CLICK on
the "show all" link is URL#2. The URL that the source code says you SHOULD
get when you click on the "show all" link is URL#1.
What's the difference between the URLs? In URL#2, the string "&part" has
been replaced with some control character or high ASCII entity.
Hopefully, I've done a better job of explaining this. If not, I'll keep
plugging away until it's clear.
To sum up: URL#1 should be identical to URL#2. It ain't.
Thanks for your time.
AB
Opera seems to dislike this usage of "&part" in the
Dines Justesen
∂ is in HTML a special char defined in HMTL4, so when O sees the &part
in the URI it replaces it with the special char. The question is whether O
should replace the &part when the semicolon is not there, as I read the
specs it shouldn't, but some people might disagree. Requireing the ; to be
there would allow people to use char entities in links, while not causing
problems at pages like the one you mention.
Dines
Dines Justesen
> in these few cases: ">", "<" and "&" and "&qout" which will be
> converted if they are followed by non-alphanumerics".
Great !!! That should solve a lot of problems, thanks.
Dines
Yngve Nysaeter Pettersen (Developer, Opera Software A/S)
We have in v4.02 changed our handling to require the "&entity;" except
in these few cases: ">", "<" and "&" and "&qout" which will be
converted if they are followed by non-alphanumerics".
MSIE turned out to have a few exceptions on what it did convert, and
it converted a couple of (legal) codes that Netscape doesn't convert.
--
Sincerely,
Yngve N. Pettersen
********************************************************************
Web-sites: http://www.opera.com
--------------------------------------------------------------------
Yngve Nysaeter Pettersen Phone: +47 23 23 48 68
Opera Software AS Fax: +47 23 23 48 70
Waldemar Thranes gt. 86 B,
0175 OSLO, Norway Email: yn...@opera.com
********************************************************************
* Opera v4.01 out now. *
********************************************************************