Do we still need the www.?

[jbsegal]

[Poll #1249826]

Discuss in the comments, of course. :)

Comments

[xeger.livejournal.com]

What's the "right thing" ?

[gilana]

I almost never bother to type it in anymore, and I usually get to the right site. I'm not a moron, but I am lazy!

[ckd]

If Netscape had pushed for SRV record use (which they could have done by statically linking a resolver library into Navigator) instead of their stupid "recognize a magic hostname, and load-balance in the client" BS back in the 1990s, then sure, pointing your browser to "foo.com" would be the right thing to do.

But they were too busy inventing <blink>, and the opportunity was lost. Dumbasses. At least Jabber (and a few other things) started using SRV.

[n5red.livejournal.com]

Since I work in .edu space, going to .com space automatically is the wrong thing.

[tla]

Where "people are morons" is elitist shorthand for "people behave like people, and usually not like geeks". It's best all around if functionality follows usage. If you don't like the prevailing usage it's almost always your problem and not theirs.

(Plus, if people are having to type in URLs by hand at all, you want them as short as possible.)

[totient]

*.foo.com should point to (edit: or into a range owned by) the firewall, which should be smart enough to forward (and distribute load) by service.

[denonymous.livejournal.com]

I know it's hotly contested, but my opinion is that while it should be www.foo.com, as that's the service, it's one line in your webserver's config to accept foo.com on your httpd port(s) and fwd along to www.foo.com or silently DTRT. In that case, you're appeasing the morons as well as not giving up your principles.

[trysha.livejournal.com]

www is the service and foo.com is the domain

foo.com should have an MX record, there should be no need for it to have an A record at all.

*gets out her cane and shakes it*

GET OFF MY LAWN YOU WEB 2.0 WHIPPERSNAPERS!

Sincerely,
uunet!ksmith!starchild!patricia

[denonymous.livejournal.com]

Actually, I just realized that doesn't necessarily check out in DNS and I may not have any idea what I'm talking about.

Carry on.

[trysha.livejournal.com]

(and yes, this is partially sarcastic.

modern web browsers, if you do not get an A record back for an address, will try to prepend "www" for the user as a courtesy.

on the other hand, many ISPs like to poison the DNS so that invalid addresses actually resolve to some sort of search engine page, usually ad-laden.

given the latter, yes, foo.com should resolve to an address, but the webbrowser should redirect to www.foo.com

[tisiphone.livejournal.com]

www not only isn't necessary, it isn't even standard anymore. Personally, I expect that typing either "www.foo.com" or "foo.com" will do the right thing through backend server magic.

[jbsegal.livejournal.com]

... which wasn't the question... :)

[warlord-mit.livejournal.com]

I second this -- SRVs can (and should) be used.
But there's nothing that says that we cannot migrate to SRVs even now.

[dougo.livejournal.com]

For some reason I interpreted "people are morons" as "webmasters who don't put web pages at foo.com are morons". People who don't type "www." aren't morons, they're just efficient.

[lonebear.livejournal.com]

www should not be a required prefix for a domain.

[feydn2.livejournal.com]

Whatever works. Mostly, anything I enter seems to make my browser happy, and if not I can adjust fairly quickly, without becoming overly annoyed (but I'm far more intuitive and reasonable than the server gods).

[deguspice.livejournal.com]

What bugs me is foo.com websites that fail )as in returns an error or points you at a blank or useless webpage) instead of moving you over to their beloved www.foo.com

[tisiphone.livejournal.com]

That irritates me too, I consider it to be a configuration flaw (likewise, a site should redirect you if you type www.foo.com even if the actual www front end is at superdupersite.foo.com).

[intuition-ist.livejournal.com]

for any given 'foo' using http, foo.com and www.foo.com are usually (but not necessarily) the same destination. it is annoying (and occasionally startling) when they are not the same.

i answered both yes and no, does that make me a network wizard? ;)

[bouncingleaf.livejournal.com]

lazy > cranky

:)

[sauergeek.livejournal.com]

You win the Intartubes. Especially with the bang path.

[the-gadgetman.livejournal.com]

It really should have been web.foo.com all along anyway....

[whitebird.livejournal.com]

People should do the right thing by using www in their URL typing.

If there's no reason to not have http forwarding when bumped by something doing http port 80 at "foo.com", then having it forward to www.foo.com is courteous if nothing else.

[lakmiseiru.livejournal.com]

I almost misread it as that myself, and had to do a couple of rereads to double check... and I'm still not sure of my own opinions on the matter :^)

[nakor.livejournal.com]

What happened to info.foo.com ?

[luckylefty.livejournal.com]

I clicked "no, other", but I really meant "yes other". The important thing is not to save the 4 bytes, but to save typing by the user.

Aren't those that insist that the user type www.foo.com being inconsistent? Shouldn't they insist that the user type http://www.foo.com?

You have the wrong appendix, actually

[(Anonymous)]

http:// is extraneous

Either use SRV records (hah) to discover the right server for the domain or we go with the traditional "www" -- note that in most cases it's actually the browser helpfully slamming the "www" on the front (and the "com" on the end), not the domain's DNS administrators putting an A record on the domain label.

[pekmez.livejournal.com]

hmm. I think not having to type it in is one thing, and I like that my browser will fill it in for me if I want to be lazy and I know I'm going to a . But I don't believe I'm typing a URL in that case, I believe my browser is magically completing the URL for me!

I want it to work, but the actual URL still ought to have the www, the service is still www, the server had still better be named www.foo.com, etc.

[sauergeek.livejournal.com]

For those who think the service is www, it isn't. The service is http. Or https. The moniker www for both (when both are present) is rather strange already.

And there is more to the Internet than the Web. There's FTP, there's IRC, there's mail (another weird one, as the service is SMTP or SMTPS, not mail, and there are specialty records for mail). All of these usually end up with suitably named hosts or load-balanced front ends.

Until we get general SRV records (to replace MX), keep naming hosts for the service they provide. The Web is just not that special; http://foo.com/ should never work absent a SRV record.

[lionofgod.livejournal.com]

I think this is something *browsers* should do, as a courtesy; and it's not so much the 4b saved as the repeated typing of mostly-meaningless characters that is one more thing aggravating my RSI. Every little bit helps.

I don't think this is something that should be done on the back end unless you're really doing smarter checking, since there *are* other services than HTTP in the world and I do use many of them.

[yesthattom.livejournal.com]

Small sites:

foo.com is the A record
www.foo.com is a CNAME to foo.com
both foo.com and www.foo.com work the same.

more geeky sites:
same as above but www.foo.com redirects to foo.com

big sites with lots of load balancers:
whatever the hell they want.

[ron_newman]

http://SomervilleMA.gov does NOT do the right thing. It does nothing at all, in fact: the web server accepts my HTTP connection and hangs forever, never returning a reply.