HTML "entities" in Opera running on WinVaster (J) Options

[Peter Evans]

Over the weekend, I created http://www.i.hosei.ac.jp/~evans/places/brighton/index.html and the seven pages linked along the top of that page. Though I'd had plenty of exercise over the last year making other people's sites rather less awful, it had been at least a year since I'd made my own pages from scratch; I thought I should do it before I got too rusty. (One oversight I know about: the pointless title attributes to most of the img tags. Feel free to criticize any thing else.) Why do it by halves? I masochistically got DOCTYPE to specify the ISO DTD and got all the pages to validate (of course!) and to display as expected on:

Firefox (Kubuntu, Mac OS X, Win2k)
Safari (Mac OS X)
Konqueror (Kubuntu)
Opera 6.03 (Win2K)
MSIE 5.5 (Win2K)
MSIE latest (WinVaster)
Opera latest (Mac OS X)

but not on

Opera latest (WinVaster)

How not? Well. . . .

Because curly quotation marks are very slightly prettier than the ASCII ones, I used them. Because I wrote the site on TSE Pro (published 1997 and designed for the "ANSI" character set) on Win2K, I wasn't able to stick these curly quotes in directly but instead had to use "’" and so forth (a waste of bytes, to be sure). But because I used these entities, there shouldn't be any character encoding "issues" , Shirley.

So anyway, the CSS lists "Trebuchet MS" at or near the top of a list of fonts to be used for everything. In the office, we have two pairs of computers running Japanese-interface Vaster and Japanese-interface MSIE. One pair has no alternative to MSIE; MSIE shows the pages well (with Trebuchet MS). The other pair also have English-interface Firefox and Opera. On these, Firefox shows the pages well (in Trebuchet MS); however, while it puts the rest of the pages in Trebuchet MS, Opera seems to render the quotation marks as double-byte Japanese characters, which are spaced bizarrely for English.

Um . . . what the cuff?

[Brian Chandler]

Feel free to criticize any thing else.

The thumbnails don't seem to lead to proper piccies... but also lack top and bottom margin, so glomm together.

--- about the quote thingy ---

Wait two or three hundred more years until there are enough multilingual people in the world to matter, and perhaps some of this will be done properly. Unless I'm much mistaken, your computers are in Japan, and you should therefore be reading text in Japanese. The browser is only trying to help.

fwiw, I used to see a lot of "Japanese quotes in English" gappiness, but don't seem to now; this is Opera/Linux and my blood group is P.

Incidentally, your page appears in a distinctive font, but I don't know if it is Trebuchet-sans-MS, or something else. There doesn't seem to be a way of finding out what font the browser is actually using except by peering at squiggles and comparing. (hmm... actually it looks to be T-avec-MS. Don't know where that came from.)

[Peter Evans]

The font should be "myriad web" else "trebuchet ms" else "yudit" else sans-serif

The pics on the top page aren't linked, and neither is the pic on the "more" page. I think all the other pics are linked to bigger alternatives. What's stuck together with what?

[Brian Chandler]

The pics on the top page aren't linked, and neither is the pic on the "more" page. I think all the other pics are linked to bigger alternatives. What's stuck together with what?

Under the bit of text about Brighton is a row of thumbnails that flow, and are spaced sideways by 3-4 pixels of space, but are not spaced vertically at all.

[Peter Evans]

Ah, them. They're not supposed to link anywhere.

I'm surprised to hear that they're spaced sideways but not vertically. In which browser do they so appear? More importantly, are the first dozen or so photos in the "people" page spaced in the same way?

[Brian Chandler]

Linux 2 / Opera 9.23

Yes the people ones are the same - images are abutted vertically. I can't see any margin (or padding) specified from a glance at your style sheet...

[Peter Evans]

Linux 2 / Opera 9.23

Hmm, ancient Opera (which I happen to have on this machine) gives equal spacing horizontally and vertically. Sparkling new Opera on Mac OS X does the same. And I hadn't noticed that sparkling new Opera did anything like this on Vaster, though then again the disgusting spacing of the quotation marks might have blinded me to this.

Oh dear. I'll attend to this a couple of hours from now.

But it wouldn't be a matter of padding or margins, it would instead be one of line-spacing -- unless I redo every drat page very radically, and I can't summon enthusiasm for that.

RANT CLASS=BEMUSED

Yes, the markup is a bit daft, really: it adheres to the (remarkably strict) letter of the ISO DTD by misapplying various tags. One way is to concatenate images in "paragraphs", each "line" of which can be centred. Another is to get around the (bizarre) prohibition of putting headers in DIVs (a prohibition that by itself was surely quite enough to condemn the DTD to obscurity) by using CSS to dolly up large, bold paragraphs to serve as headers.

This all reminds me of a fascinating web page I read some months ago on how the obsession with validation was a Bad Thing. As I hazily recall it, the author had nothing against validation in itself, let alone clean markup, but warned of trickery such as what I've indulged in and also the way that validation can leave the marker-upper with an undeserved complacency about various matters to which validation (and indeed CSS "validation") are blind: colour combinations and so forth. Certainly I've stopped the smug addition of little ticked graphics to my validating pages (visual junk, really), though I've haven't got around to removing them from (or doing anything else to) my existing pages.

/RANT

Linux 2 / Opera 9.23

Um, "Linux 2"?

[Brian Chandler]

But it wouldn't be a matter of padding or margins, it would instead be one of line-spacing -- unless I redo every drat page very radically, and I can't summon enthusiasm for that.

Really? Surely, if you give <img> a respectable margin it won't abut adjacent lines? How could you set the line spacing? Can you assume that all images are exactly the same size? Margins seem a much more robust mechanism...

... on how the obsession with validation was a Bad Thing.

Bingo. Unrelated, but... I still see the Fortran FORMAT error from 1950-something in things like:

img.i257342 {width: 257px; height: 342px;}

Is there any point in this _at all_? And some particular size has different characteristics, like being floated? This seems a travesty of sound engineering...

[Peter Evans]

Surely, if you give <img> a respectable margin it won't abut adjacent lines?

Yes. done.

I still see the Fortran FORMAT error from 1950-something in things like:

img.i257342 {width: 257px; height: 342px;}

Is there any point in this _at all_? And some particular size has different characteristics, like being floated? This seems a travesty of sound engineering...

I'm unfamiliar with Fortran, so you've lost me there.

I have a large number of images for embedding in web pages. I know the size of all of them. The overwhelming majority are in any of just four sizes. Taking it as axiomatic that browsers are being helped by being told dimensions -- a very dubious axiom, so there may indeed be no point -- it seems slightly less wasteful to have

class="img.i257342"

many times and this CSS specification, than to have

width="257" height="342"

many times with no CSS specification. What's wrong with the engineering?

[Darin McGrew]

Ideally, you wouldn't need to specify the size of images in either HTML or CSS. It's an attribute of the image itself, and the HTML or CSS shouldn't need to worry about it.

But browsers can reserve space for images before they download them if they know the size of the image. For that, I would include the size in the HTML, so the browser knows the size immediately. Otherwise it will do whatever it does for images of unknown size, and only after the CSS downloads (but before the image downloads) will it be able to reserve space for the undownloaded image.

[Peter Evans]

I agree about the ideal there. In practice, too, I suspect that today's computers -- powerful enough to let their owners [insert mindless human entertainment here] have little problem rearranging pages as images come in.

The austere ISO DTD outlaws the height and width attributes. Now, the ISO DTD is one of the great obscurities of web history, but I believe that it was made with the cooperation of people who knew what they were doing, and who therefore thought that relegating sizes to CSS would be no big deal (for browsers using CSS).

Back to the original question. Turns out that whether or not I use "entities" makes no difference. Opera on those machines pops out of Trebuchet MS and into some Japanese font when asked to render "typographic" quotation marks. I don't know why. I suppose Opera has been misconfigured, but I don't know how (I did look).

[Brian Chandler]

The austere ISO DTD outlaws the height and width attributes. Now, the ISO DTD is one of the great obscurities of web history, but I believe that it was made with the cooperation of people who knew what they were doing, and who therefore thought that relegating sizes to CSS would be no big deal (for browsers using CSS).

Why on earth would you imagine ISO has a clue about anything? They are essentially a private organization, who will do anything for money. Witness the fatuous "ISO large numbers" plastered on signboards outside factories and suchlike, telling us they now have "quality" where before they didn't, or the promotion of Microsfot's ridiculous braindump of its Whirred format which is now notionally an "international standard".

I really have no patience with this self-flagellation stuff. The only valid reason for putting the height and width things in (since you would surely use CSS if you were actually "styling" an image) is to give the browser the pixel dimensions in advance.

Back to the original question. Turns out that whether or not I use "entities" makes no difference. Opera on those machines pops out of Trebuchet MS and into some Japanese font when asked to render "typographic" quotation marks. I don't know why. I suppose Opera has been misconfigured, but I don't know how (I did look).

The fundamental problem is that Unicode is a bit of a kludge: the idea is that any "document" (or something) is in a (single) "language" (where "language" includes some pretty odd notions, like "Danish English"), and the interpretation of many "characters" (whatever they call them) varies from "language" to "language". For our European friends, here's a diagram of how more or less the same piece of text is conceived in Latin and Chinese (that's where Japanese writing came from) typesetting (remember the Chinese invented it, btw):

240 kg (21°C)

That's 13 characters, right? But the "same" text written in Japanese is 9 characters: there's no space between numbers and units, but the parentheses include built-in space so no space characters are required at all. Here it is, with each character box enclosed in [ ].

[2][4][0][kg][ (][2][1][°C][) ]

Well, problem is your quotes are simultaneously the Latin quotations, which normally come with spaces added outside, and are rather narrow, and they are also the Japanese quotes, designed to go in a square printing cell.

So have you tried telling Opera the language?? Put three copies inside <span> tags with lang=en, ja, and zh (if that's Chinese).


BTW, uname -a gives:
Linux bowen 2.6.15-26-386 #1 PREEMPT Thu Aug 3 02:52:00 UTC 2006 i686 GNU/Linux

(Do you know how to _exit_ from the man command?)

[Peter Evans]

I believe that [the ISO DTD] was made with the cooperation of people who knew what they were doing, and who therefore thought that relegating sizes to CSS would be no big deal (for browsers using CSS).

Why on earth would you imagine ISO has a clue about anything?

Oh, for example the fact that it got Abrahamson and Price, who I believe had done a lot of work on HTML 4.0, to work on the ISO DTD. The pair then went ahead and made it, with the cooperation of Dave Raggett, another familiar name. I think these people have a clue. They certainly have more of a clue than I do. (It's even conceivable that they have more of a clue than you do.)

[ISO is] essentially a private organization, who will do anything for money. Witness the fatuous "ISO large numbers" plastered on signboards outside factories and suchlike, telling us they now have "quality" where before they didn't, or the promotion of Microsfot's ridiculous braindump of its Whirred format which is now notionally an "international standard".

"Tirred format", please.

If they'd do anything for money, how come they didn't take a bit more dosh from Mircosplot -- but why the cod dyslexia every time? Is "Microsoft" untypable? -- for going further and dumping the .od? formats while they were about it? Docx oand the rest could then be unchallenged, rulz OK. That would have stuck it to pointy-headed Norwegians, GNU-drivers, commies and other enemies of free monopolies!

I really have no patience with this self-flagellation stuff.

So don't invoke it.

So have you tried telling Opera the language?

Yes, in the HTML tag.

[Brian Chandler]

I believe that [the ISO DTD] was made with the cooperation of people who knew what they were doing, and who therefore thought that relegating sizes to CSS would be no big deal (for browsers using CSS).

Why on earth would you imagine ISO has a clue about anything?

Oh, for example the fact that it got Abrahamson and Price, who I believe had done a lot of work on HTML 4.0, to work on the ISO DTD. The pair then went ahead and made it, with the cooperation of Dave Raggett, another familiar name. I think these people have a clue. They certainly have more of a clue than I do. (It's even conceivable that they have more of a clue than you do.)

Yes, but they may have provided an extremely intelligent answer to the question ISO posed (for example), but the question may have been not a very sensible one. Perhaps there are enormous philosophical advantages to not having pixel dimensions within an <img>, but I don't see how there can be practical ones.

If [ISO]'d do anything for money, how come they didn't take a bit more dosh from Mircosplot -- but why the cod dyslexia every time? Is "Microsoft" untypable? -- for going further and dumping the .od? formats while they were about it? Docx oand the rest could then be unchallenged, rulz OK. That would have stuck it to pointy-headed Norwegians, GNU-drivers, commies and other enemies of free monopolies!

What's "cod dyslexia"? (I mean, I just throw the letters in with the degree of care you-know-who uses in producing its software...

So have you tried telling Opera the language?

Yes, in the HTML tag.

Hmm, but if you did as I suggested and put three <span>s in different languages you could test if Opera actually takes any notice at all.

Have you investigated the relation between the "language" behaviour (i.e. how Opera renders 'curly quote') and the exact font specification?