Why do we have an IMG element?

November 3, 2009 8:58 AM

Why do we have an IMG element?
posted by chunking express (84 comments total)

We do not have the image tag.
posted by srboisvert at 9:04 AM on November 3 [22 favorites]

This is true. I should have been more clear.
posted by chunking express at 9:05 AM on November 3

There is no IMG tag….like the Cabal.
posted by ShawnString at 9:05 AM on November 3

We do not have the image tag.

There is good reason for that.
posted by The Devil Tesla at 9:07 AM on November 3

Why do we have an IMG element?

Because access to porn is what moves the internet forward into the future.
posted by 1f2frfbf at 9:08 AM on November 3 [4 favorites]

IMG is not an elements. Air is an element. IMG is a collection of humors.
posted by Astro Zombie at 9:11 AM on November 3 [9 favorites]

I'd love for someone to unearth the equivalent discussion about <BLINK>.
posted by mkultra at 9:13 AM on November 3 [8 favorites]

Ha! I was just about to post this. Thank you. "The ones that win are the ones that ship," indeed.
posted by bigbigdog at 9:19 AM on November 3

Yeah... it's a good read, and particularly instructive of how the "modern" html stuff came about.
posted by ph00dz at 9:22 AM on November 3

The origin of the blink tag. "Because some engineer coded it as an easter egg."
posted by smackfu at 9:22 AM on November 3 [6 favorites]

For a moment I thought I was back in MetaFilter circa 2000, where anything that WASN'T about web design would get your hand slapped.

Good times...goooood times....
posted by briank at 9:23 AM on November 3 [1 favorite]

smackfu: "The origin of the blink tag. "Because some engineer coded it as an easter egg.""

And the beast shall come forth surrounded by a roiling cloud of vengeance. The house of the unbelievers shall be razed and they shall be scorched to the earth. Their tags shall blink until the end of days.

from The Book of Mozilla, 12:10
posted by octothorpe at 9:27 AM on November 3 [1 favorite]

this thread better get use of the img tag. and not just some weak ass admin exception. the whole thread needs to be opened up to the img tag, and how.
posted by shmegegge at 9:28 AM on November 3 [1 favorite]

Early internets was so cute.
posted by contessa at 9:30 AM on November 3 [2 favorites]

Want to have some fun? Find a graybeard and whisper “Trumpet Winsock” or “MacPPP.”

Holy shit, Trumpet Winsock. That definitely brings back some memories. And I think we had MacSLIP, not MacPPP. All hooked up with 2400 baud to Telesys, UT's dial-up system. Though when I started out I didn't bother with any fancy IP stack, just a modem terminal into Telesys, then you could telnet anywhere! And you could even still transfer files! Anybody remember Kermit, XMODEM, YMODEM, and the best one of all, ZMODEM?

Not sure if I qualify as a graybeard though.
posted by kmz at 9:33 AM on November 3 [6 favorites]

oh god, trumpet winsock. My college made us use that. in 2000. thankfully, the next year saw campus wide broadband internet finally make an appearance, but jesus christ. trumpet winsock was the worst thing I've ever had to put up with, internet-wise.
posted by shmegegge at 9:37 AM on November 3 [1 favorite]

I'm old enough to remember how cool, for a brief moment, Trumpet Winsock was. An IP address?!? For your Windows machine?!? Wow!! Ultimately, of course, it was yet another Windows thingy that worked just exactly well enough to make you depend on it and then break your heart.
posted by whuppy at 9:45 AM on November 3

trumpet winsock was the worst thing I've ever had to put up with, internet-wise.

As a former Novell Netware tech, I think you led a charmed life.
posted by Bathtub Bobsled at 9:45 AM on November 3 [4 favorites]

I remember installing Windows 3.11, which I don't think was even sold, just to get the official Microsoft TCP/IP support.
posted by smackfu at 9:47 AM on November 3 [2 favorites]

This is making me all strangely nostalgic. I was in kindergarten when that discussion took place.
posted by Xany at 9:47 AM on November 3

Cause its really boring to describe this awesome picture of a cat screaming at a baby to you.
posted by The Whelk at 9:49 AM on November 3 [1 favorite]

this thread better get use of the img tag. and not just some weak ass admin exception. the whole thread needs to be opened up to the img tag, and ho


posted by fuq at 9:50 AM on November 3

What I love in Mark's summary is how simple all this early standardization stuff was. A few smart guys (and a punk kid at NCSA) just chatting about "you know what would be cool? Let's do X". And in a year or so they built the foundation of the Web, one that stands up pretty well 16 years later.
posted by Nelson at 9:50 AM on November 3 [1 favorite]

mkultra: "I'd love for someone to unearth the equivalent discussion about <BLINK>."

Good luck with that. I think it was a series of grunts accompanied by slung feces.
posted by brundlefly at 10:01 AM on November 3

The words themselves are all Greek to me, but that is a gorgeous blog.
Century gothic?
posted by Flashman at 10:03 AM on November 3

As a former Novell Netware tech, I think you led a charmed life.

I worked with what I think was the last guy in Canada still doing Banyan Vines training once upon a time.

And yes, networking on DOS was hell, but trying to configure a SunOS 4.3 box from scratch was no picnic either. And I too remember how aweosme PPP was versus SLIP because it negotiated an IP address automagically for you. Sweet, sweet modem pools of my youth.
posted by GuyZero at 10:10 AM on November 3

Metafilter: where it's always 1992
posted by mazola at 10:12 AM on November 3 [3 favorites]

trumpet winsock was the worst thing I've ever had to put up with, internet-wise.

The storyteller turns off all the lights in the server room, and illuminated only by the steady glow of thousands of small LEDs, he shines a flashlight on his face, and uses his best Bela Lugosi voice...

...and there, at the bottom of the memo from the CIO's office was the following satanic incantation, written in the syllabary of the darkest demons, recorded in blood red ink, gently smoking and wriggling, almost as if they were alive: LOTUS NOTES.

There is only the collected screams of the assembled group as everyone runs out of the room, for the light...
posted by 1f2frfbf at 10:17 AM on November 3 [8 favorites]

Find a graybeard and whisper “Trumpet Winsock” or “MacPPP.”

hey, how did he know my beard is getting gray?
posted by lodurr at 10:26 AM on November 3

Thank god they made it self closing or else I can easily imagine the nightmare of

<>Lotsa text<>

as a hacky way of backgrounding and layering of images.
posted by wcfields at 10:28 AM on November 3

I used Trumpet Winsock and MacPPP. I found a gray whisker in my beard a couple weeks ago. Until now I didn't know the two were related.
posted by zsazsa at 10:32 AM on November 3 [1 favorite]

Century gothic?

Looks more like Gill Sans to me.
posted by Rat Spatula at 10:36 AM on November 3

Why are people guessing about what font he uses? View Source to the rescue!

body{font:normal medium 'Gill Sans','Gill Sans MT','Goudy Bookletter 1911','Linux Libertine O','Liberation Serif',Candara,serif;margin:1.75em auto;width:40em;line-height:1.75;word-spacing:0.1em}
posted by chunking express at 10:39 AM on November 3 [1 favorite]

body{font:normal medium 'Gill Sans','Gill Sans MT','Goudy Bookletter 1911','Linux Libertine O','Liberation Serif',Candara,serif;margin:1.75em auto;width:40em;line-height:1.75;word-spacing:0.1em}

Die, MS Word, die!

I still have the PC that I first ran Trumpet Winsock on. It's part of a bookshelf somewhere in the basement, I think.
posted by Artful Codger at 10:53 AM on November 3

huh, that's interesting: his font cascade is basically backward: He's starting the the common font and cascading to more uncommon ones. With that cascade sequence, Gill Sans MT, Goudy Bookletter 1911, Linux Libertine O, Liberation Serif and Candara will most likely never get used, even on systems that have them. Pilgrim knows better; he must not have coded the stylesheet.
posted by lodurr at 10:58 AM on November 3

Psssst...some of us graybeards are beard-free and always have been...

What was the difference between XModem, YModem, and ZModem anyway?
posted by Asparagirl at 11:04 AM on November 3

I find it hard to believe he didn't code his sites style sheet. Do all Linux systems ship with Gill Sans? BSD? Etc?
posted by chunking express at 11:05 AM on November 3

lodurr, I'm not sure what's considered backwards about that. He probably likes Gill Sans better than the others, so he puts it in the font list first. CSS font lists shouldn't be in uncommon to common order, they should be in the order that you like the fonts.
posted by zsazsa at 11:06 AM on November 3 [5 favorites]

What was the difference between XModem, YModem, and ZModem anyway?

Sometimes I squint at the web and imagine it in Hercules Graphic Card Yellow with CGA emulation artifacts.

I miss the line by line reveal of squashed yellow interlaced porn.

Someone should make a site for this.
posted by srboisvert at 11:21 AM on November 3

XModem, YModem, and ZModem

ZModem was way faster, as I remember things. If you go to Wikipedia you can read up on the actual differences between them all. Sounds like ZModem works in a fashion similar to how TCP sends packets.
posted by chunking express at 11:26 AM on November 3

What was the difference between XModem, YModem, and ZModem anyway?

XModem was one of the first ever protocols for sending files over a wire, and worked, but barely. It sends a tiny chunk of data, waits for an acknowledgement from the receiver, then repeats the process. It is painfully slow and efficient, and on top of that had very few features (it couldn't transmit filenames, or multiple files per session, and couldn't even ensure the resulting file was the same length as the original).

YModem was an attempt to patch some of the holes, but wasn't ambitious enough.

ZModem was a modern (for the time) rethink of what the most efficient, reliable, flexible, user-freindly protocol would look like, and fill in all of the holes. Crucially, it was designed around the sender continuously streaming data until informed of a problem by the receiver (e.g. a request for retransmission of a corrupt segment), which made the best possible use of bandwidth.
posted by cillit bang at 11:42 AM on November 3 [7 favorites]

Needs more blink.
posted by msbutah at 11:47 AM on November 3

And if I remember correctly all of them were basically heaven compared to Kermit.
posted by You Can't Tip a Buick at 12:21 PM on November 3

Incidentally, I think it's kind of cute that there was all that discussion and Andreessen still ended up going with his initial proposal. Why do we have IMG? Because he added it.
posted by smackfu at 12:47 PM on November 3

CSS font lists shouldn't be in uncommon to common order, they should be in the order that you like the fonts.

Well, yes, but if the fonts on the tail-end of the list will never be displayed to a user, what's the point of having them there? And I've never seen a system that had Gill Sans MT but not Gill Sans, so I'm pretty sure Gill Sans MT has a vanishingly small probability of being used. It's either backwards or it's got too many fonts in it; either way, it surprises me in Mark Pilgrim's stylesheet.

and oh dear god, kermit was a nightmare. I started using modems during the transition from ymodem to zmodem, so when I started doing support a couple of years later and had to learn kermit, it was like having icicles jabbed into my temples. what the hell.

the whole xmodem/ymodem/zmodem thing reminds me that at one point I actually read a very thick book about how to get online. From the days when we didn't say "online". I don't even remember what it was we did say. Talk about frangible memory....
posted by lodurr at 1:09 PM on November 3

recorded in blood red ink, gently smoking and wriggling, almost as if they were alive: LOTUS NOTES

Between you, me, and the blue... there is a fortune 500 company that was still using Lotus Notes up until three weeks ago (the migration took 2 months).

If you want to really push stereotypes, one of the things we had run into is the plants south of the Mason-Dixon line refused to update. Summary quote: "Microsoft has too many bugs in their software. We don't want any of their crap messing up our systems." Even when we politely pointed out that Lotus Notes only runs on Windows platforms and everything they were using was NT based, they still shook their head and, in a very Southern drawl "We don't have time to mess with the new crap."
posted by Bathtub Bobsled at 1:11 PM on November 3

Incidentally, I think it's kind of cute that there was all that discussion and Andreessen still ended up going with his initial proposal

Isn't that pretty much just how he operated? the discussion was just cover; even in the earliest messages he was basically saying 'this is what I'm going to do.' He didn't actually broach the possibility that he'd consider doing it any other way.
posted by lodurr at 1:12 PM on November 3

would this be the appropriate thread to wax nostalgic about Mosaic 3.0, which had full-page scaling, tabbed browsing, was fast as hell, and never made it out of beta? I seem to recall it was a good couple of years before anything (that being Opera) approached its feature-set. Colors were a bit different, though, if I recall....
posted by lodurr at 1:14 PM on November 3

Between you, me, and the blue... there is a fortune 500 company that was still using Lotus Notes up until three weeks ago (the migration took 2 months).

There's a Fortune 14 company that I'm pretty sure is still using Lotus Notes.
posted by kmz at 1:18 PM on November 3 [1 favorite]

Between you, me, and the blue... there is a fortune 500 company that was still using Lotus Notes up until three weeks ago (the migration took 2 months).

Yeah, seriously, Fortune 500 companies are why Notes exists:

FORTUNE Global 500: More than half of the global 100 corporations use Lotus Notes and Domino software
9 of the top 10 aerospace and defense corporations
8 of the top 10 automotive corporations
8 of the top 10 banks
4 of the top 5 consumer products corporations
7 of the top 10 electronics corporations
8 of the top 10 insurance corporations
8 of the top 12 pharmaceutical corporations
9 of the top 12 telecommunications corporations

Source: FORTUNE 500, April 30, 2007; FORTUNE Global 500, July 24, 2006.

posted by smackfu at 1:23 PM on November 3

Hmm, I was curious about the fonts as well. He has a post on the redesign and a post about font-embedding: Fuck the Foundries.

I'm still sort of puzzled.
posted by cowbellemoo at 1:28 PM on November 3

The stylesheet for his book on HTML5 is pretty awesome, though.
posted by cowbellemoo at 2:00 PM on November 3

Well, yes, but if the fonts on the tail-end of the list will never be displayed to a user, what's the point of having them there?

Just in case.
posted by JHarris at 2:22 PM on November 3

ZModem was a modern (for the time) rethink of what the most efficient, reliable, flexible, user-freindly protocol would look like, and fill in all of the holes.

But no where near as fun as using Super-ZModem. With built in sysop chat, multiplayer connect 4 and tetris to play while you download from Rusty and Eddie's
posted by MiltonRandKalman at 2:24 PM on November 3

So, why didn't they ever create an embed tag for pulling html in from other files?
posted by Chuckles at 3:27 PM on November 3

well, chuckles, you could sort of do that with the embed tag in NN4. I don't think they actually meant it to work that way, though.
posted by lodurr at 3:42 PM on November 3

current date: Tuesday, September 5908th, 3:42pm PST.
posted by Justinian at 3:44 PM on November 3 [1 favorite]

So, why didn't they ever create an embed tag for pulling html in from other files?

That's an interesting and important question. Hyperlinks were only one feature in early hypertext systems such as eternal vapourware Xanadu. Transclusion was another important property, the ability to pull bits of stuff in from other sites. The <img> tag is essentially transclusion, as are Javascript scripts, Flash embeds, etc. But there's still no simple way to transclude some text on the Web.
posted by Nelson at 3:57 PM on November 3

So, why didn't they ever create an embed tag for pulling html in from other files?

I've always wondered that myself. embed and object and iframe all seem useful enough for what they are.... but they don't give you arbitrary text/html insertable into the flow of the document at any point, which is what you can do with the image element. You have to dynamically process or assemble your HTML documents on the back end in order to do this. Or use javascript.

It's really kindof weird that you can insert images and scripting into arbitrary document locations... but not text.
posted by weston at 3:58 PM on November 3 [2 favorites]

As a former Novell Netware tech, I think you led a charmed life.

As a former LANTastic * user, I think you all have no idea of the boundless suffering that networks can cause.

And all of you dissing Trumpet Winsock can FUCK OFF AND DIE. There was nothing wrong with it that you couldn't blame squarely on Windows 3.1. Of course, us OS/2 users didn't need a Winsock client to connect to TCP/IP. Good times ...

* especially fun when used over 10BASE2 coax!
posted by me & my monkey at 4:32 PM on November 3

Why do we have IMG? Because he added it.

You'd be surprised at how much of the internet this narrative applies.
posted by dhartung at 6:31 PM on November 3

Between you, me, and the blue... there is a fortune 500 company that was still using Lotus Notes up until three weeks ago (the migration took 2 months).

Actually, the last four fortune 500 companies and the last two government departments that I worked for _all_ still use lotus notes. It's more common than you'd think in big business....
posted by jaymzjulian at 7:48 PM on November 3

ZModem? Bah. The true players rolled with New Punter.
posted by jquinby at 8:23 PM on November 3

YModem-G was faster than your fancy-pants ZModem if you had a 16550 and error-correcting modem.
posted by jewzilla at 8:41 PM on November 3

Oh, 10-Base-2. I remember thee well. At the time I was working in an office (also running Lantastic) where the whole thing was connected via dodgy coax. When one of the crimps would start to go south, the whole network would go down. (If you wanted to add a new workstation, you had to temporarily bring the whole network down in order to insert a new T in the loop.) It was my job to walk around to each office and wiggle the T at the back of the machine every time the network mysteriously stopped working.

In case you wonder why anyone would put up with this mess, it was because you could network a number of PCs without a hub or switch, which at this point in time could not be found in the $5 bargain bin as today.
posted by Rhomboid at 8:50 PM on November 3

I like the sound of that elusive element. On the face of it, it sounds like a better idea than EMBED and OBJECT and the general multiline proprietary mess they spew out.
posted by Harry at 3:03 AM on November 4

Well, that was supposed to be "INCLUDE element".
posted by Harry at 3:04 AM on November 4

You'd be surprised at how much of the internet this narrative applies.

Eh. I don't think it would surprise most of us, at least not on this thread.

OTOH, to folks like the ones on that discussion thread, it would seem only right and proper that that's the case.
posted by lodurr at 5:41 AM on November 4

I think the reason there's not an INCLUDE tag is that it would be a recipe for disaster. Can you imagine how many broken pages there would be out there? It's bad enough when people create XSSI includes with an unclosed tag, imagine if anybody could HTML-INCLUDE anything anywhere.

Unfortunately for richer hypertext, HTML proved a fantastic bargain in the Pareto sweepstakes. The genius of HTML was that it was good enough to do what was needed without getting hung up on what theorists thought it ought to do.
posted by lodurr at 5:47 AM on November 4

One argument against an INCLUDE tag is that you have no idea what tags are in the included doc. So you can't really render any of the page until the whole page and its includes are downloaded. In the days of 2400 baud modems, progressive page load/display was a big deal so that might have been a deal killer. (Doing it server-side avoids this problem.)
posted by smackfu at 5:56 AM on November 4

LANtastic, oh dear god. I'd repressed those memories, so very deep and so well...

You win, you vicious monster, now make it stop, please for the love of all that is holy, make it stop....

*breaks down sobbing in a corner*
posted by 1f2frfbf at 6:28 AM on November 4

Our company still uses Notes and I die a little every day when I have to use it.

Case in point, as far as I know you can't perform a search on anything other than the header info.
posted by jpdoane at 9:41 AM on November 4

Index your mail and you can search on anything.
posted by smackfu at 10:21 AM on November 4

I think the reason there's not an INCLUDE tag is that it would be a recipe for disaster. Can you imagine how many broken pages there would be out there? It's bad enough when people create XSSI includes with an unclosed tag, imagine if anybody could HTML-INCLUDE anything anywhere.

Disaster? Show me a little broken document icon there, or an error message, or nothing at all. No more of a disaster than a broken hyperlink.

One argument against an INCLUDE tag is that you have no idea what tags are in the included doc. So you can't really render any of the page until the whole page and its includes are downloaded.

I think this is a better argument, but you could probably deal with this by assuming that any included fragment should be open-close balanced and having the browser do the manipulation to ensure that it is (just like browsers do a bit of document-rewriting in a bunch of other situations). Then the rendering problem is pretty much reduced to the same situation you've got with image data.
posted by weston at 5:08 PM on November 4

Actually, on furthre reading, in one of the linked mails Andreessen does cover why they won't do INCLUDE:

We're not prepared to support INCLUDE/EMBED at this point; it raises a number of nasty issues that are quite separate from the idea of inlined images. For example, what happens if one EMBEDS a document that in turn EMBEDS the first document? Oops. Aside from this, I'm not sure I see the point in allowing arbitrary EMBED's for things like chunks of texts: this is a hypertext system, after all, and it ought to be possible to get the functional effect of an EMBED by using an ordinary link. Right?

Wrong!
posted by smackfu at 5:16 PM on November 4

We're not prepared to support INCLUDE/EMBED at this point; it raises a number of nasty issues that are quite separate from the idea of inlined images. For example, what happens if one EMBEDS a document that in turn EMBEDS the first document? Oops.

I can see not wanting to do this in 1993 when you're working by yourself and just getting inline multimedia into your browser, but by 1996, we had frames, which have pretty much the same problems (plus a few more to boot).
posted by weston at 9:05 PM on November 4

How to annoy a Z-modem user: send ^XB00 to his terminal.
posted by scalefree at 12:21 AM on November 5

As a former LANTastic * user, I think you all have no idea of the boundless suffering that networks can cause.

On location at a state government that shall remain nameless, the staff thought my team couldn't hack them because their network was largely token ring. Good thing I'd raided the local surplus PC supply shop & picked up some LANtastic PCMCIA cards a few weeks earlier.
posted by scalefree at 12:35 AM on November 5

weston, it occurs to me that you might be joking, so take this straight answer for what it's worth, but: frames don't really do the same thing as an include. Includes are commonly used to just insert text (which may be code or the output of code) into the text stream. This was (ok, is) an especially common use of SSI on the server-side. Done it many times myself. You have none of the issues related to a frame or an iframe.

That's how I read what was being discussed in the thread. There are lots of other obvious problems with it. E.g., unless you're always embedding whole documents, you've got to have a way to bound what's being brought in from another document and a way to address it. (That would be really neat, BTW, and I'd love to be able to do it. I don't remember encountering it in my pre-HTML readins on hypertext, but I'm sure it must have been there, it's such an obvious "want.")
posted by lodurr at 3:38 AM on November 5

Disaster? Show me a little broken document icon there, or an error message, or nothing at all. No more of a disaster than a broken hyperlink.

Yes, if you parse the include for tag-closure (including extra tag closures), you can have a reasonable assurance that it won't break the closure of any tags in the current page. But that's not even done by browsers now; why would we expect them to have done it then?
posted by lodurr at 3:41 AM on November 5

If I were designing a text inclusion system, I'd invent a new MIME type that was an HTML fragment, not a full HTML document. Ie: no html tag, no head, no body. Just a bunch of HTML markup, you know, the stuff you care about. Then you'd convince everyone to publish the bits of text you want as HTML fragments. Wikipedia, for instance, could do this very nicely. http://en.wikipedia.org/wiki/Metafilter is the main Metafilter article: http://en.wikipedia.org/wiki/Metafilter:References could be the URL for the fragment for the "references" section. Or whatever, each site could have its own convention for serving fragments. Also you could make some parsing kludge for importing a full HTML document inside another ("ignore all tags that don't make sense").

I'm confident it could be made to work. I'm less confident it would have much value. What are the use cases for text transclusion? We don't even transclude images much other than from our own site (ie: few cross-site image tags). Transclusion gives some flexibility for page construction, letting a complex web page assemble itself from little pieces. But these days we do that just fine via JSON or XML and Javascript.
posted by Nelson at 7:23 AM on November 5

actually, there's tons of image, video, audio and flash transclusion.

i think the standard answer for why we don't see it with text is that there isn't a plain-HTML way to do it. It's done all the time on the back-end via XML, and a lot on the front end via JavaScript document.write. I think it's reasonable to suppose that if it were easy it would be done more often; how much more often is anyone's guess.

Use case: you pretty much gave one already, in being able to transclude some atomic bit from Wikipedia. But I would get finer-grained than that and offer a few other examples:

Microformat chunks: An Address or a Vcard would be an ideal thing to transclude. Or the content of an ID3 tag.

Quotes, in the sense of quotable-quotes and stuff you want people to transclude.

Self-contained or at least autonomous knowledge fragments, like explanations of some simple procedure that you don't want to have to maintain on your own. ("To download this image in Firefox using the Wombaterooni extension...")
posted by lodurr at 7:35 AM on November 5

weston, it occurs to me that you might be joking, so take this straight answer for what it's worth, but: frames don't really do the same thing as an include... You have none of the issues related to a frame or an iframe.

Not joking, but definitely aware of the difference. What I'm responding to is Andreesen's apparent concern "ooo, scary, recursive includes!" Frames definitely had problem as early as 1996, when I remember writing programs to generate pathological cases that would lock up Netscape. Which would sometimes lock up the machine. Good times.

unless you're always embedding whole documents, you've got to have a way to bound what's being brought in from another document and a way to address it... Yes, if you parse the include for tag-closure (including extra tag closures), you can have a reasonable assurance that it won't break the closure of any tags in the current page. But that's not even done by browsers now; why would we expect them to have done it then?

Browsers have been parsing for implied tag closure for as long as I've been using them; when I first learned HTML, almost nobody closed their <p> or <li> tags. In fact, my recollection is that a lot of people ignored closing tags for table markup as well. There's also other munging and reworking that happens now if you stick elements that aren't supposed to be inside one another. These problems aren't quite as involved as what we're talking about, but it shows there's some precedent, and I don't think the whole banana is that much of a bigger hurdle.

I think the big reason we've never gotten it isn't that it's particularly hard, it's that by the time anybody would have gotten around to it, the other mechanisms (js, SSIs, any dynamic server tech) existed and most people that would've thought about or needed it just went down that road rather than making it an issue.
posted by weston at 9:39 AM on November 5

but parsing for implied tag closure is not the same as parsing for broken block-element tagsets, which they did NOT do (at least not in a very useful way): viz the many blank pages I've had to troubleshoot over the years that were caused by a single missing "</td>".

As far as recursive framesets: I never locked it up, but i used to do that sometimes for fun when I was editing the one production site I ever worked on that used frames. (Where by "fun" I don't have teh expectation that anyone will understand what I mean. Or, frankly, that I would if I were back there right now.)
posted by lodurr at 11:00 AM on November 5