On 04/11/2007, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
WAKom stop.
WAKomEncoded startOn: 8080
So maybe all the Pier packages don't use the same UTF/WideStrings
default ?
I am always using WAKom, this means the model will contain UTF-8
Strings only. If you use one of the other Kom servers your model will
contain strings with a different encoding.
Well, with WAKom (or whichever is automatically launched from the
universes package), it wasn't working. I had just created a pier-blog
with an ellipsis (… alt-semicolumn on a Mac qwerty) in its
description. After stopping WAKom and starting WaKomEncoded, I could
see the blog page. Since then I replaced the ellipsis with three
periods and I can see the page with WAKom too.
Still with WAKom, I can type accented characters in a blog entry and
pier will accept them but they display as "unknown characters" in
Safari and question marks in Firefox. If I type an ellipsis I get the
partial output (seems reliable).
Back in WAKomEncoded, if I type accented characters, I get an internal
server error, with either "subscript out of bounds", or "Invalid UTF-8
input" as an explanation, depending on which and haw many accented
characters I input. But wait, if I add an ellipsis with the accented
characters, all works perfectly!
/me scratches head
Note that in all cases the page is marked as UTF 8 in the xml header,
so I suppose Safari sends UTF 8 in the form? Do the WAKom* variants do
the recoding to whatever UTF or WideStrings they try to use
internally?
Pier does not care how the strings are encoded in its
model, it only
has to be consistent. Also some web servers have (well known) problems
when the encoding is incompatible.
I'm connecting directly to Squeak for the moment, no proxy or anything.
--
Damien Pollet
type less, do more [ | ]
http://typo.cdlm.fasmz.org