Wakaba (@waka) profile picture
Osaka, Japan
Joined April 2007
data:text/xml,<!DOCTYPE a [<!ENTITY foo PUBLIC "-//W3C//DTD XHTML 1.1//EN" "">]><a>&foo;</a> &foo; is expanded to the empty string...
2
5
closed a five-years-old bug!
1
<i><b><u></i><isindex>
1
XML is difficult :-<
1
Wow, document.evaluate("/child::node()",document).iterateNext()==document.doctype is true!?
Interestingly, string-length() returns # of 16-bit code units (!= # of chars) in Firefox and Chrome, violating w3.org/TR/xpath/#strings ...
It's really waste of time to remember random string like w3.org/XML/1998/namespace ...
2
Idempotency...
1
More than 100,000 test cases are not enough to understand how browsers parse URLs...
It's far more complicated...
Compatibility is more important than consistency, I guess...
Browsers' canonicalization of mailto: URL is very simple. Good!
there are many choices, all of them are broken in different ways. sigh...
URL processing rules is back to wa1 spec, so i no longer have to refer svn diff for them. good :)
":" in authority is very, very ugly
chrome and ie both normalize ipv6 address in url, but handle ipv4-address-embedded ipv6 addresses differently...
chrome and ie normalize hexadecimal/octal/decimal ipv4 address to canonical dotted-decimal form; other browsers do not...
how browsers parse, resolve, and canonicalize urls is very interesting...