Honey, what you see is what you get, and if you've found the real thing it's the best thing yet.
--The Dramatics
TIP: If you must use a WYSIWYG editor, at least try to clean up after it. |
If you insist on using one of those "WYSIWYG" ("What You See Is What You Get"... though what you see might well not be what users of other systems or browsers might get) editors that shield you from true HTML coding, you should know about some of the junk they tend to insert into your documents. This stuff is very familiar to anyone who's tried to clean up the mess left behind by running a web site through one of those site-mangling editors.
In the main menu of this site, I refer to WYSIWYG editors as "not very well housebroken." This implicit comparison of these programs to dogs is actually not particularly fair to the canine population; in fact, dogs are quite fussy about where they poop, while WYSIWYG editors don't appear to show this degree of discretion.
Here are a few things to look for during such a cleanup job (using a regular text editor, since the WYSIWYG editors can block the user from seeing, and fixing, the HTML code itself).
Watch out for inserted tags like:
<META
HTTP-EQUIV="Content-type" CONTENT="text/html;
charset=iso-8859-1">
Some of the WYSIWYG editors stick this at the start of each document. What the above line means is "The character set for this document is ISO-8859-1." But that's already the default character set by HTML standards. Thus, there's no need for such a line as this unless you're using a different character set, like for foreign-language documents. When the above line is included, some browsers clear the screen after loading the document and re-draw it (presumably in the newly-set character set, same as the old one), producing an annoying flicker effect. Get rid of this line, and your web pages become less of a user annoyance. (If you really do need to set a different character set, it would be much better to do it by configuring the server software to send the proper "charset" identifier in the document MIME type, but you probably only have access to that if you're the system administrator.)
You may also want to remove the META tags inserted by almost all editors that advertise which editor was used. Do you really want to announce to the world (or, at least, the HTML buffs who view the code of your pages) that you use one of those goofy editors? (Actually, if you're planning on giving your web pages to another web developer eventually to fix up the mess your editor caused, leaving in this line may be desirable so the other developer knows which particular quirks to look out for!)
But do leave in the META tags giving descriptions and keywords for search engines; see my description elsewhere. Despite the importance of those tags, some editors may make it difficult for you to insert them, while they go and insert all sorts of unnecessary or even destructive tags themselves.
Some editors will stick in all sorts of bizarre things that
have no use, and may even mess up your layouts in some browsers.
For instance, Netscape Navigator Gold likes to pepper your
document with . That's a
"non-breaking space," a special kind of space character
that doesn't word-wrap (and, though this isn't part of the
official standard, also doesn't generally collapse if more than
one is used in a row, so it can be used to add extra horizontal
whitespace). Netscape's editor seems to think very often that the
user wants to have an extra non-breaking, non-collapsing space,
for instance at the end of each text line. Maybe it first
converts the carriage returns to spaces in its initial parsing of
the document, then decides that these need to be "hard
spaces" and so inserts .
That's the best I can figure out the possible "thought
process" it goes through, but the result is to stick those
silly, unnecessary characters all over the place. They make the
document bigger in disk space and transfer time than it needs to
be, make it harder to understand and maintain in a regular text
editor with all this extra junk in it, and on occasion, these
extra spaces will even find their way to a spot that screws up
the layout, like at the beginning of a table row making it fail
to line up with the other rows.
You can go through your document and get rid of those funny space characters, and then deal with the possible ruining of your layout if some of them actually were necessary to make things line up.
Other wasteful things editors will insert include formatting tags applied to empty spaces, like:
<CENTER><DIV
ALIGN=CENTER><P ALIGN=CENTER><FONT
FACE="Arial,Helvetica"><FONT
SIZE="+1"><FONT
COLOR="red"><B> </B></FONT></FONT></FONT></P></DIV></CENTER>
This whole big mess of code serves only to insert a blank
paragraph for vertical spacing, accomplishable via <P></P>.
All the other tags are useless. They're added because the editors
are so dumb that if you have stuff like font settings enabled
they insist on adding them even to blank spaces. The editors are
also pretty dumb about failing to collapse redundant tags. Even
if the various font changes above were actually needed
to make sure that blank space was rendered correctly, you could
have done it with:
<P ALIGN=CENTER><FONT
FACE="Arial,Helvetica" SIZE="+1"
COLOR="red"><B> </B></FONT></P>
Note how the three different centering tags were reduced to an attribute of the single paragraph tag, and the three different font settings were made into attributes of one FONT tag. This produces a shorter, cleaner, more logical piece of code, showing the advantages of coding by hand instead of using some silly editor!
WYSIWYG editors will also stick some weirdo oddball attributes
into tags, like NATURALSIZEFLAG in the IMG
tag. This is not part of any HTML spec that I am aware of, is not
used by any browser that I know of, and I haven't the slightest
idea what it means, but some of those editors like to stick it
in.
Watch the syntax of the HTML tags generated by your editor. Many of them do improper things like failing to put quotes around attribute values that require them (any that include characters other than letters, numbers, and hyphens; for instance, "100%" and "+1"), failing to properly nest blocks (you should use closing tags in the reverse order of the opening tags), etc.
Sometimes some of the editors will insert code that works in
one browser but doesn't have the same effect on another.
Microsoft FrontPage has the tendency to produce code that's fine
for MS Internet Explorer but not so nice in Netscape. One example
is how FrontPage will change all centering commands to <P
ALIGN=CENTER> even if you originally inserted them
as <CENTER>. While, in most cases,
using an ALIGN attribute is more logical and elegant than the
awkward CENTER tag, there are places where the only reliable way
to center something is to use <CENTER>,
like for centering tables and form submission buttons. These
aren't considered part of a paragraph by Netscape (since a
paragraph, by the HTML specs, can contain only character-level
content such as text or inline images), so they won't be centered
by <P ALIGN=CENTER>. Internet Explorer
seems to be rather more liberal in what it'll consider to be part
of a paragraph, and FrontPage follows this attitude by using only
the paragraph centering attribute even when editing a page that
was created using a different tag. The result is a page with some
elements not properly centered in Netscape, and every time you
try to fix it, FrontPage will "helpfully" keep changing
it back.
If you don't watch out, your editor might insert links and graphics in the form of hardcoded URLs of files on your hard disk instead of the proper relative URLs that work on the actual web server. Look for such code as:
<IMG
SRC="file://C|/www/sitefiles/mypic.GIF" WIDTH=200
HEIGHT=300 ALT="Picture of me">
All such URLs starting with "file:" are invalid for use on public web sites, since they point at your local hard drive and nobody else has access to that but you. Editors will tend to stick in such URLs quite often (sometimes using operating-system-specific characters like backslashes instead of the proper forward slashes, making the syntax not even valid).
This is a hard error to track down as long as you're using a WYSIWYG editor to hide your actual HTML code from you, and you're only viewing your site from your own system, on which the hard-drive references work fine. You'll need to view your pages in a normal text editor to see which URLs are screwed up and fix them.
Elsewhere I give reasons
why you should link back to your main home page with <A
HREF="./"> instead of explicitly naming
your index page (e.g., index.html), and
similarly, to link to subdirectory indices with the directory
name instead of the explicit index filename. No WYSIWYG editor
that I've encountered yet does links in this manner; instead,
they all seem to force these links to be in the form of explicit
filenames, and may even change your properly-formed
directory-name links to a hardcoded "index" reference
without you knowing it.
By default, many WYSIWYG editors fill in sections of code that are useful to users in "non-standard" browsing situations with un-helpful content. I'm referring to things like the ALT text in images, and the NOFRAMES section of framed sites. These are useful to users of older browsers, text-only browsers, text-readers for blind users, and, don't forget, the search-engine indexing robots. The use of meaningful content in these sections is important to the overall accessibility of your site, but too many users let these things be filled in by their editors in a "default" manner. This results in the NOFRAMES section saying nothing but "This site requires frames and your browser does not support them", and the ALT text of images containing something pointless like the filename and byte size of the image. See my sections on frames and images to find out what you should be putting there.
Since WYSIWYG editors let you add stuff to your site with a point-and-click interface, they encourage you to pay little attention to your filenames and directory structures. If you don't watch out, you'll wind up with a tangled mess with graphics scattered over lots of different directories, filenames with random mixtures of uppercase and lowercase letters, and other chaotic things you'd probably not do if you did it by hand. (See my comments on directory structures.)
All WYSIWYG editors tend to use lots of tables for page layout. Some, most notably NetObjects Fusion, go further and generate tables from hell, with enormous numbers of rows and columns with specific, finely-set pixel widths and heights in order to (attempt to) achieve a highly precise layout. The result, in addition to a site that is probably highly sensitive to the user's browser version and screen width and unlikely to degrade well on any "nonstandard" platform, is incredibly convoluted HTML code that's almost impossible to maintain in any other manner than in the editor that created it in the first place. So if you expect to ever switch editors or fine-tune your pages by hand, you had better not start designing them in an editor that generates this sort of code.
Even in the less-Byzantine tables generated by a more "run-of-the-PageMill" editor, watch out for bad style such as hardcoded table pixel widths that won't work well when users have screen widths different from your own.
Since WYSIWYG editors create HTML code based on the visual
appearance of the document rather than its logical structure
(even though HTML is a markup language intended to represent
logical structure), it's likely to fail to "get the
point" of your document's structure and mark it up
logically. You may well end up with your main headers being
marked up with font sizes instead of H1 and
other header tags, while other parts of your document may have
non-header content marked up with H tags
because the editor decided that was the best way to do a font
effect. Your paragraphs may be broken with <BR><BR>
instead of <P>. The result may well
look the same as a logically-marked-up document (at least in the
browser versions the editor is coding for), but text-mode
browsers, blind speech readers, search engine robots, and other
programs that try to interpret your logical structure in a manner
other than the graphical rendering you're used to, are likely to
get lost. (For instance, the content of your <H1>
header is often weighted heavily in search-engine indexing, while
the same header marked up merely with a font size will probably
be treated as much-less-important plain text.)
So, do you really want to use those editors? They may seem to be saving you some time when you're first starting to set up a site, but in the long run, you'll be spending a lot more time fixing up the problems they cause.
Here are some of the specific things I know about that certain individual editors do:
ALT=""
attribute so they're suppressed in text-mode browsers
like Lynx, leading to "[INLINE][INLINE][INLINE]"
all over the place. <H1>
as is logical, but a FONT SIZE tag
instead. When you choose a block quote, it gets rendered
not by the proper BLOCKQUOTE element,
but by a misused UL tag. The creators
of NOF go way beyond the call of duty in avoiding any
semblance of logicality in their tag usage. (See my
discussion of physical and logical
tags.) SIZE="+1".
IMG SRC
references accordingly. Thus, if the site had the
graphics neatly placed in a separate directory to begin
with, once this editor got hold of it they'll end up
strewn all over the place in all the directories you ever
edited HTML files in. You can wind up with lots of copies
of the same graphic wasting space on your server. Even
CGI-generated graphics like counters tend to get changed
into static graphics by this editor. CENTER tags to P
ALIGN=CENTER whether you like it or not, and
even in circumstances where the latter won't work in some
browsers. </TABLE> tag of a table
before the opening <TABLE>
tag, so that it was closing a table that hadn't opened
yet, and then never actually closing the real table once
it started. Netscape can't cope with that sort of bogus
code, though Microsoft Internet Explorer is somewhat more
tolerant. The author swears that the code is
exactly as Front Page generated it and he didn't mess
around with it using any other program, though I have
trouble believing that even Microsoft would generate HTML
this bad. <HEAD> section
the following line:<meta http-equiv="Content-Type"
content="text/html; charset=windows-1252">This page was first created 29 Nov 1997, and was last modified 23 Jun 1998.