Test of Beta 4B

[MikeGale]

I think that's a great idea.

[thacker abcdefg]

Gentlemen--

I am not so sure that an overly long list of 'web-safe' fonts is a good thing.

'Web-safe' implies two things:

1. The typeface has been designed for the font rendering engine of the various browsers.

2. The font is widely distributed as a device font across various operating platforms.

[Albert Wiersch]

Gentlemen--

I am not so sure that an overly long list of 'web-safe' fonts is a good thing.

'Web-safe' implies two things:

1. The typeface has been designed for the font rendering engine of the various browsers.

2. The font is widely distributed as a device font across various operating platforms.

Hello,

It seems there is some confusion.

In v9.0 BETA 5, there are only 20 typeface names recognized as "web-safe". The fonts in this thread are (mostly) not considered web-safe.

[thacker abcdefg]

Thanks for clarifying that, Albert.

[MikeGale]

I personally don't use the idea websafe in my thinking.

I construct font family cascades based on my best guess of what the user audience is seeing.

This changes, fortunately slowly. It depends on your exact user base (do they use Office 2007, Vista or a Unix flavour to an unusual degree? which you can seldom measure in advance).

The settings for things like

Code: Select all

monospace
sans-serif
serif
cursive
fantasy
default

can be reset to varying degrees in different browsers (and no doubt in the registry). A user could put them all to Impact if he wanted in some browsers, which would look interesting!! They're always there but what you get is not completely reliable.

(There are some additional complexities which you can fortunately ignore in practice. Like font substitute settings in Windows (which can be very useful if you know what you're doing).)

Ignoring the alternate symbol sets like Webdings, Symbol or Wingdings. Here's a crude estimate of the prevalence of some fonts on the three platforms. (The numbers are percents in the order Windows/Mac/Unix.)

Code: Select all

Andale Mono 3/85/48
Arial 96/96/62
Arial Black 98/96/53
Arial Narrow 87/91/0
Comic Sans MS 96/92/50
Courier ?/96/75
Courier New 96/92/62
Georgia 91/93/51
Helvetica 7/96/58
Impact 96/88/53
Times 4/87/50
Times New Roman 85/90/55
Trebuchet MS 95/93/50
Verdana 97/94/54

I'm not even going into the issues with Courier, which I haven't seen written up anywhere!

Looking at these you can see that nothing is really safe (given that the base statistics are not perfect either).

(The list is a combination of what you see in CSE and from elsewhere.)

So unfortunately you need a cascade and you need your eyes open, unless you can rely on something like EOT (downloadable fonts) which seems to work across browsers on Windows.

[thacker abcdefg]

Gale--

EOT is and has been supported by Microsoft browsers since 1996. With the CS3 EOT proposal supported by both Microsoft and Adobe, typography within Web documents will change if EOT or an offshoot is adopted across browsers.

Font embedding [a misnomer that is generally applied in reference to EOT] and font linking [supported only by Safari, currently], both accomplished by the @font-face construct, will make font-family specifications more critical particularly for fixed width layouts.

Web safe typography, in my view, will become important and the bulk of that may reside in how font foundries expand upon embed permissions and if they market packages of web safe typefaces.

[MikeGale]

You've put your finger on the key point with availability.

I haven't pursued EOT much because I can't practically make reasonable fonts myself and there are so few acceptable quality embeddable fonts out there.

In my last test on several browsers on Windows all the browsers worked with @font-face EOT's. This surprised me.

There is a peculiar note I've seen that seems to suggest that CSS 2.1 was abandoning support for EOT. I've never bothered to track down what was going on. Do you have any idea?

(As for fixed width layout I personally think that's a really odd thing to do. For me web pages are about flexibility and using the power of the medium, not crippling it by artificially imposing the limitations of paper.)

[thacker abcdefg]

Gale--

EOT was recently submitted as a spec recommendation to the W3C for CSS 3.0. http://www.w3.org/Submission/EOT.

Use, creation and testing of Microsoft's EOT is problematic. Microsoft's WEFT tool needs a major overall and that has been acknowledged by Microsoft.

A font management application is a necessity. EOT can only be created when the typeface resides within the Windows Fonts folder, i.e. device font. Once created, the typeface needs to be removed from the Windows Fonts folder and then managed by the font management application, e.g. activated/disabled.

The reason you thought EOT was working across all tested browsers was because the primary font resided in your Windows Font folder, thus a device font, that was referenced via your font-family CSS construct.

Beyond Microsoft IE, the @font-face construct is only supported by Safari v3.1 as spec'd by CSS 2.1 -- i.e. a uri to a 'raw' font on the server.

Regarding fluid vs fixed, much of that is dependent upon project, I believe. Fluid is the trend and firmly supported by staunch standards advocates. Lately, for the hell of it, I have been looking at a combination of both for designed content that will accommodate the entire view port, maintain aspect ratio and eliminate all scroll bars.

[MikeGale]

Thanks for that Thacker.

I didn't check what WEFT was doing under the hood, when I decided it was too hard to get fonts to use with it. Now I have a better idea.

[MikeGale]

For those here who are not familiar with WEFT and EOT.

It's a technology for putting any font (if permitted), onto a web page. It involves a preparation stage during which you pick the glyphs (characters) and web site on which it will work. This produces an EOT file. You reference that file in your stylesheet and the font shows on the page.

In my original tests a copy of that downloaded file was on my machine, as an installed font. It thus showed up in all browsers. (I don't know how it got there, it wasn't deliberate.)

I just re-tested the technology, using Firefox, Safari, Opera and IE, after uninstalling the offending font.

This time it performed as expected. Only worked in IE (not Safari though it's a 3.1 version). The font did not get installed onto the machine. In other words it seemed to work as expected this time around.

Thacker, I dont put myself in the standards fold.

(The flow idea long preceded the coining of the term standards. I've seen standards advocates who promote development techniques and beliefs that I disagree with.)

[MikeGale]

I've had a quick look at some Safari @font-face notes. On the surface it looks as though the implementation expects a ttf file. (I have not researched on my own machine.)

UPDATE

I ran some crude tests. Safari picks up a distributable TrueType (ttf) font, IE doesn't. IE (7) picks up an Embedded Open Type (EOT) font, Safari doesn't.

So we have a classic and avoidable web situation. Here is a tool that is genuinely useful and currently can't be used unless you get complicated and have one of each (which might not work!!).

Isn't life great!