W3C home > Mailing lists > Public > www-html@w3.org > December 2003

Re: a recommendation - Received on Thursday, 18 April 2002 14:01:42 EDT

From: Ernest Cline <ernestcline@mindspring.com>
Date: Wed, 3 Dec 2003 07:58:59 -0500
Message-ID: <410-220031233125859984@mindspring.com>
To: "Christian Wolfgang Hujer" <Christian.Hujer@itcqis.com>, "Fastpitch Central - Bill" <bill@fastpitchcentral.com>, www-html@w3.org
Cc: www-xml-linking-comments@w3.org




> [Original Message]
> From: Christian Wolfgang Hujer <Christian.Hujer@itcqis.com>
>
> Hi,
>
>
> I didn't read the thread, though have some annotations:
>
> Am Mittwoch, 3. Dezember 2003 08:00 schrieb Fastpitch Central - Bill:
> > Frank Tobin said:
> >
> > It's not smart to have the content layer start making up ad-hoc
solutions
> > for problems in the code/protocol layer.  Furthermore, how would a
client
> > even receive the page that has this information in it, if it's not
> > addressable with an A record?  Catch 22.
> I agree, the IP address still works - as long as the base hasn't been
tampered 
> with unwisely. <base href="/groups/" /> works fine, while <base 
> href="http://myserver/groups/" /> will break down if the DNS fails.
> But nearly the whole internet breaks down when the DNS fails... *eg*
> Wise those running their own forwarding caching DNS servers for they keep
> the internet more redundant and reduce the traffic.
>
> > I for one could search around and type the "dotnum" into the client
browser
> > to get the website started.  And, if my <base ...> tag modification
were in
> > place I could then surf the site without any problems.
> Well, they're obsolete now.
> <base/> has died.
> Long live @xml:base!
> ;-)
>
>
> > If the <base . . .> tag suggestion I made was accepted then folks would
at
> > least have an option.  Businesses and government agencies could have
> > their dotnum posted in emergencies.  They could give out the dotnum
> > whenever appropriate in those, hopefully rare, emergency situations.
> Perhaps before continuing any discussion about <base/>, you should
> become familiar with the XML Base recommendation and see whether the
> problem you're talking of still exists in XML Base.

I did go back and read the proposal. The idea is interesting, but I fail
to see why a backup base URI should be restricted to just dotnums
if references cannot be resolved according to the primary base URI .
Suppose for example a site which has mirrors on other sites.  It would
be useful if all of the mirrors could also be referred to in the base.

xml:base does not provide this level of redundancy.  If it were provided
that would probably be the place to do it, but while the benefits are easy
to perceive, I also see potential problems.
1) More complicated code needed to resolve URIs.
2) A security problem at one base becomes a problem for all.
3) When and how often does a user agent need to pick a base?
4) Maintaining synchronicity of the bases.
5) In the case of a dotnum level redundant base URI, how does
   this affect load-leveling servers that pass off requests against one
   name to multiple IP's?

I've sent a copy of this to www-xml-linking-comments@w3.org
as I believe that is the appropriate list to discuss this idea.
At this point while its interesting, I'm not certain if the benefits
outweigh the potential problems, and while I know enough to
perceive some of the problems, I don't know enough to be
able to judge how serious they are.
Received on Wednesday, 3 December 2003 07:59:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:59 GMT