on Mon, 27 May 2002 14:31:04 GMT, [EMAIL PROTECTED] (Martin A. Hansen)
wrote:
> my nice cgi forms works beautiful under konq on linux. but the
> windows machine generates ^M at the and of lines? this makes eval
> go fubar? how can i prevent this???
There seems to be a lot of confusion about the line separators in
<textarea> fields in an HMTL-form (if that's what you are talking
about). In the HTML 3.2 reference spec at
<http://www.w3.org/TR/REC-html32>
one reads about <textarea>:
It is recommended that user agents canonicalize line endings to
CR, LF (ASCII decimal 13, 10) when submitting the field's
contents.
It seems that konq doesn't follow this recommendation, while the
Microsoft IE 5.5 does.
Further confusion arises when representing these special characters in
various ways: decimal, octal, hexadecimal, etc. The following table
(Perl notation) illustrates this:
Name Decimal Octal Hexadecimal ControlChar
CR 13 \015 \x0D \cM
LF 10 \012 \x0A \cJ
Then, to confuse matters further, the "\n", aka 'newline' sequence
comes in, which happens to be different on different platforms:
Platform Newline
Unix/Linux \cJ
Mac \cM
DOS/Windows \cM\cJ
So if you're on a Unix platform, and you get <textarea> data from a
HTML 3.2 compliant browser, you're essentially hosed. You get DOS-style
line endings, while you want Unix-style ones. (And maybe some Mac
browsers return <textarea> results with \cM line endings - I don't
know).
In my opinion, to be on the safe side, you will always have to perform
both the following substitutions:
s/\cM\cJ/\n/g;
s/\cM/\n/g;
when you receive data from a <textarea> form field. You may even have
to do the reverse substitutions when you want to include default text.
--
felix
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
