FONT FACE considered harmful

Here is an illustration of the warning
http://ppewww.ph.gla.ac.uk/~flavell/...onts.html#dont
that you should not specify a typeface when you have characters
outside West European Latin-1 on your page.

The reference test page is
http://www.unics.uni-hannover.de/nht...emp/arial.html

This http://www.unics.uni-hannover.de/nht...face-arial.gif
is the horrible result in Mozilla 1.3 on Mac OS 9.1 when I do
"allow documents to use other fonts". You can see that some
[non-Latin-1] characters are displayed in Chicago instead of Arial.

This http://www.unics.uni-hannover.de/nht...ce-default.gif
is the result when I disallow to use other fonts, i.e. using the
browsers default typeface (here: Palatino).

Andreas Prilop wrote:

This http://www.unics.uni-hannover.de/nht...face-arial.gif
is the horrible result in Mozilla 1.3 on Mac OS 9.1 when I do
"allow documents to use other fonts". You can see that some
[non-Latin-1] characters are displayed in Chicago instead of Arial.

This is probably because your Arial font doesn't contain the
necessary glyphs. So Mozilla plays smart and at least shows the
right character for you. I see nothing unusual, nothing wrong... ?

The default font an user has chosen doesn't necessary contain glyphs
for all the Unicode characters (a HTML4 document use). So even if an
author doesn't specify a font face there may appear characters
rendered with different font from the default the user has chosen.

This is related to a user question who tried to display the greek
omega letter like this:

<FONT face="Symbol">W</FONT>

But why in the world one would want to display the 'W' letter like
the greek omega letter... only just because the "Symbol" font
doesn't contain 'W' glyph.

--
Stanimir

On Tue, 7 Oct 2003, Stanimir Stamenkov wrote:

Andreas Prilop wrote:

This http://www.unics.uni-hannover.de/nht...face-arial.gif
is the horrible result in Mozilla 1.3 on Mac OS 9.1 when I do
"allow documents to use other fonts". You can see that some
[non-Latin-1] characters are displayed in Chicago instead of Arial.

This is probably because your Arial font doesn't contain the
necessary glyphs.

That _is_ Andreas's point, after all (but it could be worse: in some
situations, WinIE would just play dumb and present placeholder glyphs,
instead of looking for some other font).

So Mozilla plays smart and at least shows the right character for
you.

Indeed; but it's not a pretty sight.

I see nothing unusual, nothing wrong... ?

As Andreas rightly says, it's cosmetically a mess.

The default font an user has chosen doesn't necessary contain glyphs
for all the Unicode characters (a HTML4 document use). So even if an
author doesn't specify a font face there may appear characters
rendered with different font from the default the user has chosen.

If they take some minimal interest in what they're doing, one might
expect them to be using a font available on their platform which
covers the language script(s) which they're accustomed to reading.

They're in a better position to know what those fonts are than some
arbitrary author who might not be familiar with their platform at all.

This is related to a user question who tried to display the greek
omega letter like this:

FONT face="Symbol">W</FONT

We've been here before, many times. But it's a fundamentally
different point from the one which Andreas was making.

Alan J. Flavell wrote:

We've been here before, many times. But it's a fundamentally
different point from the one which Andreas was making.

O.k. Seems I haven't got it, sorry - probably it is because of my
bad english.

But anyway, I don't think specifying font family is such fundamental
evil (if I got it right this time).

--
Stanimir

Stanimir Stamenkov:

Alan J. Flavell wrote:

We've been here before, many times. But it's a fundamentally
different point from the one which Andreas was making.

O.k. Seems I haven't got it, sorry - probably it is because of my bad
english.

But anyway, I don't think specifying font family is such fundamental
evil (if I got it right this time).

Not a fundamental evil, perhaps, but it's still not advisable.

Most of the time people specify fonts hoping that it might make the page
_look better_. But if the character repertoire is wide the actual result
could be that the page looks just as ugly as Andreas showed us (or
worse). So if your striving for nice looks, you'd better not specify any
fonts in such cases.

And as Alan pointed out, the actual result can be even worse, in some
brain-dead browsers, where some characters actually might not be
displayed at all, although they could have been, if the author had taken
the good advice.

--
Bertilo Wennergren <bertilow (AT) gmx (DOT) net> <http://www.bertilow.com>

On Tue, 7 Oct 2003, Stanimir Stamenkov wrote:

But anyway, I don't think specifying font family is such fundamental
evil (if I got it right this time).

Maybe it's not evil - but what's the point in specifying a typeface
(which is always for cosmetic reasons) when the results can be so
horrible? Is there any advantage at all for the others, i.e. those
readers who get _all_ letters in Arial, with the specification
"font-family: Arial, sans-serif"?

On Tue, 7 Oct 2003, Stanimir Stamenkov wrote:

This is related to a user question who tried to display the greek
omega letter like this:

FONT face="Symbol">W</FONT

Alan J. Flavell wrote:

We've been here before, many times. But it's a fundamentally
different point from the one which Andreas was making.

O.k. Seems I haven't got it, sorry - probably it is because of my
bad english.

The issue which is being discussed above is the fact that the "W"
character in HTML is the "W" character, and nothing else; the Omega
character is a totally different character, at a different character
code position, and is in no way a cosmetic variant of the "W"
character. (The example that I use in my web page[1] on the topic is
"q" versus "theta", but the principle is the same.)

It's certainly an important principle, and one on which I think you
would agree with me (and I know that Andreas would). But the point in
that case is that the (misguided) author of the above markup is trying
to fool the browser into using a font whose character arrangement
differs fundamentally from the normal one.

But anyway, I don't think specifying font family is such fundamental
evil (if I got it right this time).

Well, the font face = symbol/dingbats/webdings etc. thing in HTML, and
its counterpart in CSS, surely _can_ be categorised as "evil"; but it
wasn't the issue that Andreas had raised on this thread.

Andreas, on this thread, was talking about cosmetics. The author has
presumably wanted to specify a font with the intention of *improving*
the cosmetic appearance, but has succeeded in making it much, much,
worse.

all the best

[1] http://ppewww.ph.gla.ac.uk/~flavell/...e-harmful.html

But be sure to read the classic
http://babel.alis.com:8080/web_ml/html/fontface.html

Andreas Prilop wrote:

On Tue, 7 Oct 2003, Stanimir Stamenkov wrote:

But anyway, I don't think specifying font family is such fundamental
evil (if I got it right this time).

Maybe it's not evil - but what's the point in specifying a typeface
(which is always for cosmetic reasons) when the results can be so
horrible?

The reason for specifying things for just cosmetic reasons is just
that - the cosmetic reasons (author intentions and aesthetic look).
If it wasn't the case there wouldn't be CSS, IMHO.

If a specific design is good or bad - it's another (very relative)
issue.

Is there any advantage at all for the others, i.e. those
readers who get _all_ letters in Arial, with the specification
"font-family: Arial, sans-serif"?

My default font is Arial. Generally I don't like specifying custom
fonts not because it alters the type face so much but it alters the
preffered font size. (which in generall is a function of the
corresponding font face 'font-size-adjust' and how narrow the font
face is).

--
Stanimir

Stanimir Stamenkov <s7an10 (AT) netscape (DOT) net> wrote:

X-Accept-Language: bg, en-us

I noticed that you have Bulgarian as first language. Here is
http://www.unics.uni-hannover.de/nht...netscape48.gif
how Netscape 4.80 displays my test page
http://www.unics.uni-hannover.de/nhtcapri/cyrillic.win
*when I add* style="font-family: Arial, sans-serif" to the source
text. Well, you may call Netscape 4 broken in the first place.
Nevertheless, I think it's also the author's fault to specify
"Arial" for non-Latin-1 text.

Alan J. Flavell wrote:

It's certainly an important principle, and one on which I think you
would agree with me (and I know that Andreas would).

Yes, I agree with you on all points.

Andreas, on this thread, was talking about cosmetics. The author has
presumably wanted to specify a font with the intention of *improving*
the cosmetic appearance, but has succeeded in making it much, much,
worse.

For this and many other cases the HTML spec
<http://www.w3.org/TR/html401/present/styles.html#h-14.3.1> states:

User agents should also allow users to disable the author's style
sheets entirely, in which case the user agent must not apply any
persistent or alternate style sheets.

--
Stanimir