:Archive Of November 2001:
Monday, November 26, 2001 - 8:35 AM -
I was stretching a stylesheet around some content last night and found another bug. Simple stylesheet, basic bug. CSS1.
This has got to stop.
CSS1 was written in 1996. Clearly bug reports don't work. No doubt there are some good engineers working on this, but they're not getting anywhere. And they're not going to get anywhere unless the executives behind the three browsers decide it's a priority that their products read the standards they're made to read.
Sure, Håkon Lie is CTO of Opera. Opera fails full compliance to CSS1. Partial compliance isn't an option. We've got pages to get out the door for all web devices. We can't be screwing around with rendering bugs in these desktop graphic browsers. We've got to be able to write valid code and ship it. Today.
The WaSP is a great idea. It's an embarrassment to humanity that a diplomatic pressure group is needed to convince browser manufactures that their browsers should read the code they're supposed to read, but it's a good idea and the WaSP is doing useful work. But it's not enough.
The WaSP is only going to get anywhere if the browser company executives can hear a howling mob outside. Then they'll listen to cool heads at the table.
This is going to take being loud and hostile. Not unintelligent, and not without useful feedback in calm terms as well. But distinct, painful hostility for all deviations away from full standards compliant rendering. Keep them on the path.
It's my first few coffees in the day so I'm just filling out the idea. What we have to work with is the Web. That's our lever. Let's make it long enough. Speak out in all forums, create new ones, and get more people making their own pages so they know exactly what we're talking about. Old story; you've got to speak truth to power.
Way, way back a couple of gentlemen were discussing the trouble of getting their country, the US, to pay any attention to what it was doing to the land, water, and air we all live in. One remarked that the problem with politicians is they just follow the way the wind is blowing. Well, said the other, then we need a big wind. They formed the Sierra Club. Things aren't rosy now, but if you compare American awareness now to back then, it's a changed country. It's time we cranked up our own wind. The web is too culturally important to the entire world to back off.
Thursday, November 22, 2001 - 7:57 PM -
It's a sound night. Somafm for net radio. Tram'n'bus for remembering where you are not now.
For the eyes, app-art.org. "Software art is the name given to an emerging medium of artwork, an offshoot from the web-centric work of net.art. Artists are increasingly expressing themselves using code to produce executable applications."
This is what google looks like.
- 5:06 PM -
I still think it's weird you have to take a cd out its case to play it. Instead of just jacking the whole thing, like a floppy.
Wednesday, November 21, 2001 - 6:15 PM -
Okay. No one has said they see the bug on ALA. So. Well, whatever it is, it's clearly related to using a background color for a:hover, so for the few layouts affected the quick fix is don't use a bg color. Let me be clear that the layouts in the test are the only ones I've seen this bug in. All other CSS layouts I've done have used bg color for hover with no trouble.
I'm going to give this a few days to see if I can get a better idea for isolating it further. And perhaps we'll find someone who does see the bug in the current ALA. Meanwhile the test data stays up for everyone's reference, and I've added this handy index page should you care to bookmark.
Thanks. That really was a big help running the tests. There's just no other way with some of these bugs than to get everyone to compare the same samples at once.
- 10:13 AM -
What? Doesn't anyone see the guillotine/scrolling bug on A List Apart? This is too strange.
Well that leaves the bug data I've got so far, so I'll start with that, but I'd really like to find out who's seeing it happen on ALA to round things out.
..and my computer just got quiet. I believe a fan just died. Excuse me while I post hastily and shut down.
- 7:47 AM -
(Thanks Rory.) Opera joins the big league: here's the security flaw. In this article about it, "Jon Von Tetzchnor, chief executive of Opera, points out that Internet Explorer and Netscape browsers have been affected by similar cross-site scripting flaws in the past." Um. Which tells me Opera had plenty of advance warning to look at the problem in their browser and didn't.
Personally I don't use cookies. As a developer I understand how darn useful they are. And I figure if you're a developer and you're ignoring their dark side and forcing people to use them to view your site, you're not a very good developer. That's my two cents. My main browser has cookies off. For sites I really need to use that require cookies, I've got a copy of Netscape 4 that I use as a 'junk browser', similar to having other email accounts to handle spam.
If you're using Explorer, have a look at this. I haven't tried it, but it looks interesting, and I like that you can set security for individual pages. It seems to be a high-tech feature skin, not just a security plug-in.
Tuesday, November 20, 2001 - 2:02 PM -
Well, darn it.
We've run two test sets -yay team- and gotten at least one feedback for each Windows except NT and 95, and the results are consistent. Which really isn't what I was counting on happening.
So why are some people seeing the ie6 guillotine bug, aka scrolling bug, on ALA, but not in that second test set? Are you running ie6 right now? Click this ALA link. Do you get the bug there? Then tell me about it, please.
(And does anyone have ie6 on NT? It'd be a help if you'd run the tests below.)
Monday, November 19, 2001 - 8:33 PM -
Neato. Got results from most of the ie6/Windows OS combinations, and we all seem to be seeing the same bug. Now let's try the ALA code, which is the one that really doesn't make sense. It's a similar layout, and it has the bug, but not in exactly the same way, and it doesn't react the same for me as for some other people. If we can figure out when and where it's reacting differently, we get one step closer to the needed workaround. [Fun, huh? Yeah, I'm just a blast at parties.]
So what's OS got to do with it? Here's the second test set. Tell me what you see.
Sunday, November 18, 2001 - 6:28 PM -
Using ie6? Got a few minutes?
Here's the problem. I'm getting reports the "jumping link guillotine bug" is inconsistent. My own results are consistent, both to create the bug and to fix it, so I'm wondering if it matters which Windows OS one is using.
So here's the test set. I'd like to get feedback from others to help sort this out for all of us.
Thursday, November 15, 2001 - 1:00 PM -
BBC: The European Parliament is taking action against cookies. "Cookies form the fundamental transmission mechanic that powers the internet," said Danny Meadows-Klue, chairman of the IAB. ... The IAB claims that the move could cost British companies £187m as online adverts become less effective and advertisers re-think marketing strategies.
I'm laughing, but I shouldn't. People might buy even a fraction of that. Wow, £187m. Love to see the 'data' for that claim. And online adverts might become less effective without them? Advertisers might have to re-think? Like provide value for attention? Golly. "Powers the internet." Woo!
This is a cheap shot, but is that the man's real name? I thought a meadow clue was something you avoided stepping in.
Ahh, ad men.
Tuesday, November 13, 2001 - 5:40 PM -
That's actually pretty funny. In the detailed press release for version 6, Opera repeats their stance on the handling of bg image:
"Positioning of background images is relative to the element box, not the window. This means that in Opera, a image placed with body (background-position: center) will be roughly in the middle of a page, not in the middle of the window."
But they fixed it! Said example puts the bg image in the middle of the window where it should be, though not in agreement with their interpretation above, which is arguable depending on how you read that part of the spec.
Thank goodness. We can use positioned body bg images now. Perhaps marketing just wasn't alerted to the change.
A quick tour through Box Lesson reveals no new bugs, but also no bugs are fixed. Opera is still not fully CSS1 compliant. That's sad. Those bugs have to get fixed, and it's baffling that they won't acknowledge them.
So far, 6 is quite similar to 5.12; they've removed nothing I like. Plus you can install it to a different directory and run both at the same time. No need to risk your bookmarks and email in a beta. Rendering is improved; they fixed an odd jagginess that occurred while scrolling over bg images.
And you can finally use the spacebar to scroll! Hooray! That's the one shortcut I've missed from IE.
Great browser, btw. If you haven't tried Opera yet, try it now.
- 2:25 PM -
Opera 6 beta for PC is out. Here's hoping I can run it alongside 5.12 for testing.
Monday, November 12, 2001 - 12:19 PM -
Buzz Club. "The artists at P.S. 1 are a generation younger than the 'Superflat' crew; in some ways they live in a different world. They're online, on the phone..."
- 9:51 AM -
Update. The 'three columns with a top box all touching' example also fails in ie5mac, which is not a good thing, and by the bizarre amount of 10px, which is very disconcerting.
Again, that page layout is just going to have to be restricted to some fudging either by bg colours or by messing with text margins for a few years until MS's evil twins are behind us.
The 10px error is disturbing because the layout did work with a 20px fudge in the box margin, or a 20px fudge in the text margins. I don't know what's going on here and worry about where else it will bite us. Padding is 10px. Perhaps it's showing the same error as ie5pc but with an extra padding twist that I wasn't able to isolate before.
Saturday, November 10, 2001 - 5:31 PM -
If you've seen Willy Wonka's Chocolate Factory (or read Charlie And The Chocolate Factory), you'll remember the setting was a magnificent tour of the world's greatest and most secret candy factory.
That's what I want to do with Google. I want to see the hardware, read the logs, learn the algorithms. Find out how they really structure and organize and staff the thing. Listen to them explain their interface design strategy. How do they do it?
Nobody knows. You gotta work there. (And probably sign one heck of a NDA.)
I just wanna know. The worst thing is that if I ever do meet anyone from Google, I'll have an awful time not doing the Oompa Loompa song.
Friday, November 9, 2001 - 6:38 PM -
Runs fine in sp2. Sp1, anyone?
- 1:57 PM -
Nope. Can't be done.
To get three columns with a top box all touching, you'll always have the top of the center column sit too low in IE5 for PC, by the amount of your text top margin. I can't make a hack that doesn't trip up in IE5.5basic. If you're going to insist on that design and aren't able to use bg colors to hide the glitch, then the workaround it is to set your text top margin to zero, then build up the bottom margin to preserve your formatting.
Thanks, MS. Maybe next time you could read the spec you're implementing?
(Yeah, I don't usually put out straight flamebait, but I'm fed up with all three browser companies. This is ridiculous. We've got so much useful work to be doing and we're standing on our heads trying to make valid code work in busted browsers thanks to lazy, arrogant, incompetent, paid programmers. It's a 1996 spec.)
Anyway, I'm pretty certain the modified code will run sweetly in 5.5sp1 and sp2, but I wouldn't mind screenshots.
Now to clear my head and get back to useful work. We've got a web to build.
Thursday, November 8, 2001 - 4:34 PM -
Oh zark. Still waiting for someone to help out testing IE5.5 basic, sp1, and sp2 against each other, but I think I've found the source of the trouble mentioned before.
That page gives trouble because of IE5 and 5.5's really lurid content placement problems, detailed here. So there's a hack in the code, using the mechanism of Tantek Celik's width hack. Except somewhere between 5.5basic and sp1 or sp2, MS fixed part of the content placement problem but not the ability to read Tantek's sleight of hand. So 5.5basic doesn't need the hack info but can't see the right info in Tantek's hack. I guess the sp1 or 2 that I used when building the box lesson could see the hack, but since it displayed the rest of 5's content trouble I figured that one still existed too. Or something like that. Hard to be certain without running all three.
So at least I know where to dig, and it looks correctable. Anyone have 5.5 sp1 or sp2 on a machine that you can send me screenshots from? I'm anxious to fix this before too many more people try using that code.
- 3:09 PM -
Whee... Since I mentioned the old box lesson in a post last week, I was sent a screenshot of how the page fails spectacularly in IE6. Which didn't surprise me too much because the code is full of nasty hacks that were later refined in the current box lesson. But then I saw that the bg image of the main div appeared twice. Hello.
Turned out this was triggered by the negative bottom margin I used to get the title text to snug up together, part of one of the nasty hacks to make the page work for IE5.
Fine. We're not likely to sling much code like that. But there's been other really odd results out of IE6, where apparently unrelated items have an effect. Like a dashed border appears dotted when you go to a page. It goes to dashed when you resize the window. Dotted when you refresh. And then there's that decapitation nonsense that appeared in this blog until I stopped using a bg color in a.hover.
I'm starting to get fearful of how IE6 is put together. I hope they dump the damn thing and build on the rather nice IE5Mac for IE7.
Owen Briggs ©2000, 2006
