Reader's Comments

on Learn to Program HTML in 21 Minutes
Using HTML now is like using machine language programming was before compilers were invented. Programs, such as Claris HOMEPAGE, have put HTML out of business. You don't have to learn any makeup; just type and select options, as you would with any text editor.

-- Dr. Dermott (Red) cullen --, January 21, 1998
If you think that HTML is bad as a formatting language due to the poor quality documents that result, think about it from the web server/client point of view. It's even worse.

There is so much logic that has to go into parsing an HTML page that web programs (servers, clients, CGI scripts) have to be incredibly complex. This is error prone. Because HTML was designed as a human-readable formatting language (based upon SGML, which is the original evil) it is inherently complicated from the computer's point of view. People forget formatting tags, or use incorrect formatting tags, or type in the formatting tags using incorrect syntax. There are a million problems of this type that HTML parsers have to deal with. More fundamentally, HTML is just a stream of text and lacks any fundamental structure which makes the parsing even more difficult. Plus there are tons of ambiguities in the HTML specs coupled with archaic conventions which must be adhered to even though they are not specified fully in any spec.

HTML is just a big lose, no matter which way you look at it. I think that someone ought to come up with a new web formatting language, and develop a web server/client which can handle both HTML and this new, more intelligent markup language. Hopefully this would allow HTML to die the slow and painful death it deserves.

-- Bryan Ischo, January 30, 1998

To expand a little bit on the

discussion: HTML is an application of SGML. SGML is good. Part of the definition of HTML (in SGML) specifies: MINIMIZE OMITTAG YES SHORTTAG YES In SGML-speak, this specifies two useful features. First, any tag which could be deduced from the structure of the document may be omitted. This includes not only

, but also , , , and many others. The second feature, SHORTTAG, allows markup to be expressed more concisely, e.g., Shortened End Tags for Fun and Profit. Unfortunately, since very few Web browsers actually implement HTML as is it formally defined, these features cannot be relied upon. However, it is a Simple Matter of Programming to combine existing SGML tools with `make' and a copy of the HTML DTD to automatically process Web documents through an SGML normalizer which transforms HTML into a common subset understood by all browsers. At the same time, other advanced SGML features could also be employed to advantage, such as user-defined entities. (See, for example, The FreeBSD Project's Web pages at for an example.) For my personal Web publishing efforts, I use PSGML, an SGML mode for Emacs, which understands DTDs to a limited extent and is capable of performing most sorts of normalization automatically. It also has editing features such as an element attribute editor, so you don't have to look up whether IMG takes a SRC or an HREF attribute---it will just present the full list, and you pick the ones you want.

-- Garrett A. Wollman, April 6, 1998
I agree wholeheartedly with most of your views on website design, with all its excruciatingly slow- loading pages and unecessary gimmicks. However, your comments on graphic designers is a bit off- target. Any PROFESSIONAL designer considers users' loading-speed, bandwith, size of screen and so on, the same as would be done with any printed product (public needs first). You seem to lump together graphic designers and computer graphics people. The first will examine the audience's needs before thinking of ANY concept or design idea,this being a very important part of their training. The latter knows how to visually masturbate with Photoshop and all, with little consideration to how it'll all come out in the end. Those are responsible for most of the terrible sites you talk about. Also, you seem to forget that a well-designed (read:pretty) site will beat out in popularity any site containing the exact same information. People are used to "eye candy", and any crappy page screams "amateur" to many people. Good-looking sites, designed by experienced graphic designers, are what makes the internet so enjoyable today. As long as they load quickly :)

-- Walter Palfy, December 1, 1998
The part about people's misuse of HTML is on-target, but the bashing of HTML itself strikes a sour note.

The reason for the WWW's popularity is precisely the laxity and imperfection in HTML as opposed to SGML. It provides a low barrier to entry for the hobbyist or non-technical person. Someone feeding comma-separated lists into a database would not sustain their interest long enough to put their accumulated knowledge online. Someone doing "Save As" and finding some specialized tag set would not be able to transfer that coding to their own web page on a different topic. If browser writers had to validate and reject all non-conforming documents we'd still be on Netscape 1.0 and nobody without a CS or engineering degree would have a home page.

I've observed many struggle with SGML -- waiting 1/2 hour for a document to validate, abandoning WYSIWYG editors in favour of Notepad, typing verbose tags to convey meaningless distinctions -- this is not the kind of thing that could or should "take over the world".

Let's look at the example of "<em> preferable to <i>". Italics is a typesetter's convention to indicate emphasis. <em> is a contrived convention, takes an extra character, and conveys nothing extra. 99.9% of the population decides to italicize based on a few simple rules -- such as "is this phrase like anything else that is normally italicized".

We've heard for years, "just mark everything up with maniacal granularity and someday wonderful search engines will make life easy". Yet nobody has the inclination or funding to write search engines that separate out the 10 different reasons why one might want to emphasize a phrase. It just adds to the cognitive load of the writer. I've marked up many thousands of "program keyword" tags when everyone else was using "bold" but nobody ever wrote a search engine to reward these efforts. It's a mug's game.

The only such distinction that might be worthwhile is for book titles. One could add an automatic linking capability. Yet even then it turns out to be easier in practice to use Perl and search for a set of known strings between <i> tags and treat them as book titles, than to look for <cite> and trust that some omniscient author will encode link-target information.

-- John Russell, April 4, 1999

Add a comment