Re: Spurious characters

On Mon, 20 Aug 2007 07:09:46 -0000, Eirik Byrkjeflot Anonsen
<eirik@opera.com> wrote:

> 0 <root@mouse-potato.com> writes:
>
>> Another PITA unique to Opera (this does NOT happen with
>> konqueror).
>>
>> I see this with very few sites, but when it occurs, it
>> certainly can annoy.
>>
>> Look at this page:
>>
>>
http://js-kit.com/pop_comments.cgi?ref=http%3A%2F%2Fglobaleconomicanalysis.blogs
pot.com%2F%2F&path=http%3A%2F%2Fglobaleconomicanalysis.blogspot.com%2F2007%2
F08%2Ffutures-fireworks-and-moral-hazards.html
>>
>> Search for 835, and you should get a URL that looks like this:
>>
>>
http://www.latimes.com/business/la-fi-countrywide17aug17,0,1835165.story?coll=la
-home-center
>>
>> In konqueror, you get exactly what you see.
>>
>> But with Opera (9.22, and at least 9.21 before), if you copy
>> the text and paste it, you actually get:
>>
>>
http://www.l%E2%80%8Batimes.com/business/la-f%E2%80%8Bi-countrywide17aug17,0,1%E
2%80%8B35165.story?coll=la-home-center
>
> e2 80 8b is zero-width space.  It seems the text contains zero-width
> space characters at the given places.  I haven't checked the source of
> the page, but I would guess that this is actually what the text looks
> like.  Though there seems to be no reason to put those characters
> there.  Why they would want potential space (and possibly line breaks)
> at those places is a mystery to me.

I finally figured out the URL of the actual data. To get a more
recent example:

http://js-kit.com/comments-data.js?ref=http://globaleconomicanalysis.blogspot.co
m%2F2008%2F04%2Fglobal-recession-on-way.html&jx=0&gen=0&srt=date&
;ord=asc&sp=1&pn=2&ps0

It seems that Opera can't handle the <wbr> tag (although their use
in this application seems wrong at best). From what I can tell,
every other major browser supports this tag.

> I'm not sure what the correct fix is.  I don't feel comfortable with
> removing characters that really are in the text (assuming that they
> are there, of course).

I don't believe that Opera should replace the wbr tags with "zero-width

0 <root@mouse-potato.com> writes:

> On Mon, 20 Aug 2007 07:09:46 -0000, Eirik Byrkjeflot Anonsen
<eirik@opera.com> wrote:
>
>> 0 <root@mouse-potato.com> writes:
>>
>>> Another PITA unique to Opera (this does NOT happen with
>>> konqueror).
>>>
>>> I see this with very few sites, but when it occurs, it
>>> certainly can annoy.
>>>
>>> Look at this page:
>>>
>>>
http://js-kit.com/pop_comments.cgi?ref=http%3A%2F%2Fglobaleconomicanalysis.blogs
pot.com%2F%2F&path=http%3A%2F%2Fglobaleconomicanalysis.blogspot.com%2F2007%2
F08%2Ffutures-fireworks-and-moral-hazards.html
>>>
>>> Search for 835, and you should get a URL that looks like this:
>>>
>>>
http://www.latimes.com/business/la-fi-countrywide17aug17,0,1835165.story?coll=la
-home-center
>>>
>>> In konqueror, you get exactly what you see.
>>>
>>> But with Opera (9.22, and at least 9.21 before), if you copy
>>> the text and paste it, you actually get:
>>>
>>>
http://www.l%E2%80%8Batimes.com/business/la-f%E2%80%8Bi-countrywide17aug17,0,1%E
2%80%8B35165.story?coll=la-home-center
>>
>> e2 80 8b is zero-width space.  It seems the text contains zero-width
>> space characters at the given places.  I haven't checked the source of
>> the page, but I would guess that this is actually what the text looks
>> like.  Though there seems to be no reason to put those characters
>> there.  Why they would want potential space (and possibly line breaks)
>> at those places is a mystery to me.
>
> I finally figured out the URL of the actual data. To get a more
> recent example:
>
>
http://js-kit.com/comments-data.js?ref=http://globaleconomicanalysis.blogspot.co
m%2F2008%2F04%2Fglobal-recession-on-way.html&jx=0&gen=0&srt=date&
;ord=asc&sp=1&pn=2&ps0
>
> It seems that Opera can't handle the <wbr> tag (although their use
> in this application seems wrong at best). From what I can tell,
> every other major browser supports this tag.
>
>> I'm not sure what the correct fix is.  I don't feel comfortable with
>> removing characters that really are in the text (assuming that they
>> are there, of course).
>
> I don't believe that Opera should replace the wbr tags with
"zero-width
> spaces" as you refer to them.

I agree, we should not do that.  In fact, I suspect we don't do that.
Note that the actual html is generated by a javascript.  It wouldn't
be surprising if they actually replace the <wbr> with a zero-width
space because they believe that the browser does not support <wbr> and
they're hoping that ZWSP will work better.

I see that in my (internal build of) opera 9.50, they show up as
ordinary spaces, but do allow line breaking.  So we don't really
handle that correctly either...

However, if I "Mask as firefox", those spaces disappear.  Both from
the screen and from copy-and-paste.  So my suspicion that opera gets
different data than safari increases.