Overuse of Floats Considered Harmful definitely deserves a link. I’ve thought the same thing for a while now, and do my best to avoid using them except for things like floating pictures in columns of text. You can tell, when you try to use them for overall page layout, that they just weren’t made for it and attempting it is hackish at best. Hopefully the new layout features of CSS 3 will arrive soon, and soon enough that float usage isn’t hardcoded into authors’ minds like table layout has been.
Archive for the ‘web’ Category
Overuse of Floats Considered Harmful linkage
Thursday, December 29th, 2005Bayesian Tagging
Friday, December 2nd, 2005So why haven’t I heard of any CMSes using Bayesian filtering to automatically suggest tags for blog posts? Am I reading the wrong places? Have I heard of it, and just have a bad memory? ‘Cause this can’t be a new idea…
Copy-pasted RDF
Thursday, November 24th, 2005What percent of RDF on the web is copy-pasted?
Just curious.
Installed ComPreVal
Wednesday, November 9th, 2005I just installed ComPreVal. Hopefully it should do the same pseudo-XHTML* checking and forced preview as what I had before, and hopefully without the backslash problem I had before. Bug me if anything’s broken.
* Pseudo because it’s text/html. If this for some reason offends you you can think of it as (pseudo-X)HTML. As to why I’m publishing anything at all with a text/html MIME type and an XHTML (Transitional, ewww) doctype, it’s because that’s how the CMS came, and I don’t have the skills to fix it… yet.
(Markup) changes I would make
Monday, October 31st, 2005This is a list of markup changes I would make to languages on the web, if I were all-powerful and able to push them through, and backwards compatibility weren’t an issue:
- Specs would be written so their content models ignored elements from other namespaces. For example, when html 4 says
<!ELEMENT UL - - (LI)+ -- unordered list -->, I can do<ul><foo:bar><li>blah</li></foo:bar></ul>and all is good and well and valid. - Forms would be a separate namespace from documents. See my previous post.
- Navigation/branding/other template-type things would be their own namespace. See again my previous post.
insanddelwould have their own namespace.- New namespaces would be easy to generate, and it would be easy to make machines (partially) understand them, without needing those machines to be updated at all. See my post on this topic.
More to come… this post was mostly inspired by the ins/del thing.
Reading back over the first item… is that really what I want? And just say if authors do something like <html:em><grammar:paragraph>text text</grammar:paragraph></html:em> it’s their problem? Or do I really just want some way to say “this element doesn’t affect content models”?
Document/Presentation/Application
Wednesday, October 12th, 2005Blah, this is my third time writing my post, and I’m not really feeling writerly right now, so I’m just going to say what I need to quickly and get it over with.
Basically, the “Markup / Style / Behavior” split that everyone always talks about is wrong. The split really should be Document / Presentation / Application. Presentation is split into Styling—CSS—and for lack of a better word Branding—templates, a spec for which is yet to be developed. Application is split into UI—Forms—and (again for lack of a better word) Programming—client-side or server-side, or one as a fallback for the other.
Because of this I’m not really happy with any of the specs currently out there or currently being developed, with the one exception of CSS, which fits nicely into Presentation: Styling. I first started noticing this with WF2, which, although it’s one of the closest to being good, extremely bugged me by having the template attributes apply to all HTML elements, instead of just fieldsets like I said it should. I haven’t read the WA1 spec enough to comment well on it, but from passively reading the mailing list I get the vague impression that it fails to stay in one category in a bunch of ways. I haven’t read any of XFORMS, but from what I’ve heard of it it sounds like it has not only that problem but plenty of others.
Actually the first sentence of that last paragraph is wrong. XML is okay, since it doesn’t give any of the categories, it just standardizes markup. The Document, Style: Branding, and Application: UI categories will all be markup based, and having a standard markup is good. Javascript I’m only just becoming familiar with, but it seems like a good fit for Application: Programming (client-side). That’s not saying those don’t have other problems, just that as far as I can tell they’re not fundamentally flawed in the way I’m discussing.
Going to end this before I start rambling, since like I said I really don’t feel like writing, so much so that I’m not even going to proofread. So.
Reasoning for paragraph 2 left as an excersize for the reader. (Basically, this whole post is telling you what to think, you have to figure out why to think that.)
I’m currently getting more blog spam than email spam, so if you post a comment please also email dolphinling@delphiforums.com as it’s less likely to get lost that way.
Bug 192891 was fixed!!
Thursday, September 22nd, 2005Bug 192891 was fixed! Woohoo!
Mozilla, Virii, Slashdot
Wednesday, September 21st, 2005Argh, Slashdot got hold of the virus in the Korean contributed builds of Mozilla and Thunderbird. I was on IRC at the time they were discussing this, and I remember my first thought being “I hope slashdot doesn’t hear about this.”. And they didn’t even note the fact that those were contributed builds, not mozilla.org official ones, nor that ClamAV has now been installed, nor that the list of people with upload rights was already in the process of being reduced, to prevent just this kind of thing in the future.
*Sigh.* Time to go clarify the situation there…
Update: Slashdot seems to have gotten it into its collective head that the infected files weren’t hosted on mozilla.org at all. This is, of course, wrong, but if they’re thinking that, then I’m far too lazy to correct them.
Work
Saturday, September 10th, 2005work n.
Anything that you do that you would prefer to have already done.
(“Have” being a verb, not “have already done” being the past perfect of “do”.)
P.S.: What’s the proper way to mark this entry up?
Foxfire, Revisited
Sunday, August 28th, 2005Hey, cool, go and google for Foxfire. Three results down it says “See results for: firefox”, and links mozilla.org, firefox.com, and spreadfirefox.com. Didn’t do that when I searched for it a while ago. I’m still going to keep my little link over in the sidebar, though.
Firefox Software Update is amazing
Sunday, August 28th, 2005Holy cow. I just came upstairs from reading, and I saw a dialog: Deer Park has downloaded an important update, blah, blah, blah…
. And I clicked Yes, or whatever it was to continue—just one click, mind you—and then the standard “You have more than one tab open, are you sure you want to close”, and then it started back up and I was using a newer nightly. The only thing it didn’t do was bring me back to the same pages I was on already. It’s amazing. I mean, I know that’s what it’s supposed to do, but wow.
I wonder what it would do though if I had installed it as root? I put nightlies in my home directory for convenience, but if I hadn’t, would it fail gracefully? Would it check to make sure it could do it first, before even downloading? That would be best… I should check on it.
Why are feeds under different URLs?
Saturday, August 27th, 2005How come everyone does feeds with a different URL than the HTML-marked-up content? Taking Anne as an example, his blog is at http://annevankesteren.nl/, so shouldn’t I be able to point my feed reader at http://annevankesteren.nl/ and let Accept: headers and server magic do their stuff to get me a feed format?
Cracking
Wednesday, August 24th, 2005So apparently someone cracked into my blog and changed the front page. AFAICT they didn’t do anything else, so it was probably just a bug in an old version of wordpress. So I upgraded, hence the nice shiny new defaut layout.
Of course, now my comment validation plugin is gone. I wish they’d just sent me an email telling me to upgrade instead.
If we ignore web standards…
Sunday, August 14th, 2005Some days I just love slashdot…
[I]f we ignore web standards, then the terrorists have already won!
Bug 207028
Tuesday, August 2nd, 2005Can someone look at bug 207028 for me? I’m not sure it should really be invalid. Here a simple testcase.
Nevermind, David Baron answered it in the bug. Thanks! :-)
multipart/alternative spam
Saturday, July 30th, 2005Weird. I just got a multipart/alternative spam, and all the plaintext part said was Get a capable html e-mailer
. You’d think you’d need more brains than that to survive as a spammer…
Please Don’t Feed the Trolls.
Wednesday, July 20th, 2005You know another thing that would be good in a blog? The ability to mark a comment as “troll”. Give it a bit of styling—perhaps a cartoonly little “Please don’t feed the trolls” sign, or give people the option to hide them altogether. On the other hand, I can see where a blog owner might be tempted to abuse that. Hmm.
The Netherlands is filled with Web Geeks.
Monday, July 18th, 2005Holy Cow. I’m looking through Google Zeitgeist, and the Netherlands’ #2 query in May was “ajax”. It beat out “lindsey lohan”, “star wars”, and “xbox 360″, losing only to “paris hilton”. Is the entirety of the Netherlands a bunch of web geeks? You’d never see that here in the US…
Also, the #8 search for the Netherlands was “referendum”, so not only are they web geeks, they’re politically-minded web geeks. Wow.
And, since I can’t help but laugh at it, the #5 search for Norway was “pokemon”. Heehee…
Fixed!
Thursday, July 14th, 2005Okay, so I fixed the comment backslash thing. It required removing addslashes(...) from something, amazingly enough.
I just hope I didn’t leave myself open to that SQL Injection Attack thingy that page was talking about…
Now I’m gonna go through every single comment ever posted and remove them all. Funness.
Edit: Well, that’s… kind of nice. Apparently due to some strange misconfiguration, the backslashes are removed automatically when I go to the comment editing form. Unfortunately, it removes legitimate ones, too (unless they were escaped by a false one). I hope I didn’t break anyone’s post.
Sandboxing javascript for user comments
Sunday, July 10th, 2005It would be nice to be able to sandbox user comments so that they could include more than just plain markup. For styles this is trivial: provide a [this-comment] selector, and cut out any selector that doesn’t start with it (would fail IFF a future version of css introduced a parent selector, but even then could be easily fixed). For scripts, I’d assume it’s quite a bit less trivial. It would be really nice for us web geeks, though.
I even wish I could offer a bounty on this. It would really be worth it, I think, if it were done correctly. Unfortunately I’m too young for most things financial—credit card, paypal, etc. Oh well. Hopefully someone will find it an interesting challenge.