On Sat, 25 Sep 2004 21:47:06 +0100, Steve Benner wrote: > At 1:19 pm -0500 25/9/04, Jim Witte wrote: > > � Well, my 2004082608 (admitted, old) version of Camino displays the > above amazon.co.uk fine, including pound signs - which it codes > correctly as html elements.� Camino seems to select *no* text > encoding by default. > > > You'll find the problem is intermittent.� Some times the page will be > fine, other times it will not be.�� Try visiting one of their "new > and used" pages, though, (here's one at random: > http://www.amazon.co.uk/exec/obidos/tg/detail/offer-listing/-/B0000A9YZY/all/) > and you'll find Camino almost always gets it wrong (look down the > left side of the page). Camino has been like this since about last > December. > > �(why use the co subdomain at all? In the name of simplicity! Grr..) > > > Sorry, but that's just the way we do things in the UK.� Or do you > regard the UK as so tiny it isn't worthy of subdivision into > commercial (.co), academic (.ac) governmental (.gov) organisational > (.org) etc subdomains?� Pity we own almost half of the class B > networks in the world then... :-p > > > How about some coordinated web-dev activism?� Everyone on this list > write to the webmaster of Gameplay (and probably other sites they run > across) and tell them to either specify the text encoding in the > correct place in the <meta> tag or Content-type, or use the £ > HTML element as specified in the spec.� Even better, tell them to use > the Unicode encoding, which is probably what everything will end up > using in 10 years anyway. > > > I'd agree, but I'm having to lie down at the moment after banging my > head against several hundred webmasters already today for sloppy > coding... > > �for all .com, .us, .ca, .uk (and all UK country codes), > > > erm... there is only one UK country code:� .uk� :) > > and maybe a few other TLDs.� (.net and .org and probably .biz are > probably spread outside the 'ISO Latin' world� (I don't know which, > not being a character encoding expert) > > > > Truth is, you can pick any Western font and Camino will the render > the pound (and the numbers that follow it) without error.� The > problem, as you say, is that Camino is choosing NO encoding.� Firefox > doesn't exhibit this behaviour because it lets you choose a default > encoding to use (and defaults to ISO-8859-1 if you don't).� Section > 3.7.1 of RFC2616 (the definition of the HTTP/1.1 protocol--full > document is at http://www.ietf.org/rfc/rfc2616.txt should you desire > some light reading) has this to say on character encoding: > > When no explicit charset > �� parameter is provided by the sender, media subtypes of the "text" > �� type are defined to have a default charset value of "ISO-8859-1" when > �� received via HTTP. Data in character sets other than "ISO-8859-1" or > > �� its subsets MUST be labeled with an appropriate charset value. > > > Which is frequently interpreted (incorrectly -- see 5.2.2 of the html > specs http://www.w3.org/TR/html401/charset.html#h-5.2.2 ) by web > designers as an invitation to omit the encoding information if > ISO-8859-1 is in use.� BUT, while web designers are not free to make > this assumption, it is fair to say that web client designers SHOULD > apply this simple rule: > > any page delivered as text/html without a specified character > encoding should be treated as encoded in ISO-8859-1.� > > > > Again, note 3.4.1 of RFC2616: > > �� Some HTTP/1.0 software has interpreted a Content-Type header without > > �� charset parameter incorrectly to mean "recipient should guess." > > > So: clients shouldn't guess.� They should enforce ISO-8858-1.� If > Camino isn't doing, a bug should be raised. > > (<rant> This assumes that HTML is not succeeded by some other > protocol-mix in the future - XML/client-side XSLT, > > > Sorry, Jim...� it's already here (as well you know!) BUT... the > encoding rules for XML are much more strict and webmasters won't get > away with being sloppy once they start using it.� heh heh heh.� The > world could become a better place yet! :)
Rant rant rant :) The fact is, the other site posted about this, the one that I investigated, it setting the content type correctly, and the use of the � is completely valid for the character set that they are using, which alternately will work in the utf-8 character set since it extends off of iso-8859-1 for the most part. Anyway, my guess is that it's the linked files that are doing it. At some point soon, I'll tear it apart and reproduce the issue locally, but, unfortunately, we have a small storm bearing down on us in south Forida (http://www.nhc.noaa.gov/ftp/graphics/AT11/refresh/AL1104W+GIF/252103W.gif). I expect to report back tomorrow. Adam. -- ----------------------------------------------------------------------- Adam Randall http://www.xaren.net [EMAIL PROTECTED] AIM/iChat: blitz574 [EMAIL PROTECTED] "Macintosh users are a special case. They care passionately about the Mac OS and would rewire their own bodies to run on Mac OS X if such a thing were possible." -- Peter H. Lewis _______________________________________________ Camino mailing list [EMAIL PROTECTED] http://mozdev.org/mailman/listinfo/camino
