lynx-dev Re: msg00798.html (was: 0x2276 handling)

[Foteos Macrides]

"Leonid Pauzner" <address@hidden> wrote:
>>         The URL is: http://unicorn.us.com/testlynx.html
>Thanks.
>
>> <p>
>> <a href="http://cgi.unicorn.us.com/cgi-bin/unicorn/test-cgi?getpd=512&lg=1";>
>> http://cgi.unicorn.us.com/cgi-bin/unicorn/test-cgi?getpd=512&lg=1</a>
>> </p>
>>
>>         The "&lg=" in the query will be treated as "&lg;=" because the '='
>> is an implied terminator for the "&lg".  Alex has another paragraph in
>> that page where he changed the "&lg=" to "&lgin=". That works fine,
>> entirely, because "lgin" is not defined anywhere in Lynx as a named
>> character reference, and so the standard error recovery occurs, entirely.
>
>Following recovery issues proposed:
>
>(1) translate entities only from correct syntax which ends up with ";",
>                overwise return verbatim  (but the BIG TWO?)
>(2) do (1) only in HREF= (but we got two strings you quoted above
>                displayed differently)
>(3) do not translate &xyz= which ends up with "=" (no ";" before "=")
>
>The last variant looks more preferable.

        The last variant is still too general.  What might be good is
a more specific exception, that an '=' is not treated as an implied
terminator in the specific case of an attribute value which corresponds
to a URL.  That violates SGML principles, but in that specific case an
'=' not preceded by a ';' is highly unlikely to be intended by the
author as an implied terminator.  You want to be careful about what
you do for character reference handling, though, because i18n URLs are
not far from becoming commonplace.  The specs for them have for the
most part reached consensus in the IETF forums, and commercial
implementations have already been released.  As I've stressed, the
v2.7.2 code will handle those, but the v2.8 code will botch them.

        The weather here bears no resemblance to spring, so I'm still
at this terminal.  But I do hope you understand, Leonid, that I spent
two solid months working on the chartrans stuff in the code set which
I released in v2.7.2, and when it finally comes, I'd rather enjoy
spring in NYC than get involved again with the morass that was released
as v2.8.  Character handling is fundamental to virually everything a text
browser does.  It impacts on everything else that is being discussed
for further Lynx development.  So dealing with that morass in v2.8
certainly is important for the viability of Lynx.  But you can spend
only so long beating your head against loose canons. :)

                                Fote
-- 
Foteos Macrides (address@hidden)