IE6, Opera 5+, Netscape 4, lynx and w3m also support gzip and compress encoding. w3m even claims to support bzip2 encoding, although that's probably a bit heavyweight for this sort of thing:)
Anyway, gzipping content can easily make for an 8x size reduction on a large page, especially if there's a lot of repetition in there, e.g. lots of tables. Whether this translates to a significant speedup in browsing speed depends heavily on the size of the page and the speed of the connection; certainly on a modem, going from 80k to 10k is very noticable:)
Actually it struck me as being quite logical, especially for such an important syscall. It means you can get the options with a single pretty much atomic and very cheap call to lstat() -- otherwise you'd need at least an open()/read()/close(), and then you run into worrying about locking while the config file's open.
It did make me do a bit of a double-take when I first saw it though:)
Furthermore, as a small hint... on freebsd, it was always good to make a file called malloc.conf in the/etc directory, and put nothing but the letter H in it. Effectivly it will never make a difference due to the fact unused pages eventually get swaped out.. but it helps keep the resident size of your programs down...
Actually, malloc.conf is supposed to be a symlink to a file*name* containing options, not a regular file containing them, i.e:
ln -s 'H'/etc/malloc.conf
man malloc for more flags, and alternative ways to specify them.
StorageReview, uh, reviewed the PATA model a couple of months ago; they found it to be slightly quieter than the Barracuda IV, with largely similar performance.
Are IP's really so scarce that companies are resorting to stealing them to stay afloat?!
It's obviously time for us to migrate to IPv6, before this escalates and we have our first full scale war over netblocks.
Oh, sure, it's going to start with this sort of scuffle over a/32, but just you wait until we have our first nuclear exchanges over those last few/8's!
True, it's the choice of the UA - but until most of the browsers out there implements a feature like this, a web-developer simply can't rely on this feature being present, which means we are forced to "work around" it using pop-ups (or a completely separate page - which is more load to the server).
Logout will require an extra request either way, since logout should be mainly server side -- you don't want to ask the UA nicely if it'll please delete your session cookie or whatever, because it's well within it's rights to tell you to go jump in a lake, and may even pretend it's done so without actually doing anything. Clients can not be trusted.
Any "yes/no/cancel" dialog - the cancel basically acts as a back button, but by having a pop-up you can prevent the user having to do a complete HTTP request/response cycle, unnecessarily loading down both the server and client.
A pop-up explaining a problem with a users input in a form - it's a simple notification prompt, and requires only an acknowledgement (using a completely separate page is once again an unnecessary request to the server)
I hate the way modal dialogs are implemented in most UA's, though. They have no place in a browsing environment -- I want to continue browsing elsewhere and an always on top dialog that demands input and which steals focus SUCKS. Better would be to use JS+DOM/CSS to place an item on the page (maybe replacing the form) to do the confirmation.
Less compatible, but you can probably fairly easily get 90% of clients, with the other 10% getting a traditional extra request.
Glossary definitions, where a word, when clicked, links to a small description.
If the description is small, the title attribute is better. You can even use some DHTML to make it into a pseudopopup on click. Personally I'd just do a normal link with a glossary and let the user hit Back.
Picture or short article viewer, where a thumb-nail/abstract list is displayed on the "main" page, but each click generates a small window with the full content.
Probably the worst use. If they're articles and pictures, they're real content and I'd rather get them in the current page rather than screwing up my normal browsing reflexes and giving me a teeny browser window that may or may not even fit in with my environment.
and NOT using the antiquated HTTP method of authenication - for one, it has no way to "logout" a user
This is a user interface issue with the UA, nothing to do with HTTP authentication. There's nothing stopping browser developers adding a Logout button to the navbar when it's using HTTP auth.
And there's no reason we, as web developers, shouldn't be able to use pop-up windows for web-enabled apps.
There's plenty of reason, some of it you also get with frames -- they break the UI the user is used to by breaking Back, maybe disabling parts of the UI to make the window look "cleaner", etc.
Plus they don't work everywhere, so you're already breaking the "access anywhere" concept.
I've been doing this, although I've not advertised anywhere.
Build and set up a machine -- £50 + parts.
Troubleshoot some problem -- £15-£30+, depending on complexity.
The nice thing about this is you'll tend to get more clients the more you do, as people go around telling others about how you fixed their computer and how they should give you a call whenever they have problems.
Fast Forward -- fancy UI gadget -- if a site uses said links, the Forward button is turned into a Next button, which is nice for browsing things like search results pages and blogs.
Actually this feature doesn't use <link> tags. I think it just takes a smart guess at what "next" should be.
Yup, you're right. It seems to look for <link rel="next">, <a rel="next">, and <a>next</a>.
This explains why hugely respected accessibility expert Mark Pilgrim slated the MS site redesign
Yes. Microsoft's web team leaves quite a lot to be desired. But then, so do most other so called "professional" web developers who wouldn't know a standard if it jumped up and down screaming "I'm a standard! I'm a standard!".
Tantek Çelik [tantek.com] (who's site kicks major ass BTW)
Tantek's website is awful. It does at least degrade well when I turn off CSS, but it's not very nice to use. He even has a splash page ffs:)
but MS doesn't seem to want to listen most of the time...
But Tantek is Microsoft. So is Tim Lacy, and Chris Wilson, and Ed Tecot, and Laurie Anna Kaplan, and David Meltzer, and Stephen Waters, and Scott Isaacs, and.. well, you get the point. Microsoft are not a single entity out to be incompetent and ignorant, it's a whole bunch of people, mostly incompetent and ignorant. Just like any other random collection of 50,000 people;)
Improved CSS support -- CSS menus now work pretty much as expected, overflow: scroll works better, and numerous other fixes.
Opera now has a password manager! Both HTTP auth and login forms can be saved and filled in automatically later. "Wand" is a bit of a cheesy name for it, though:)
Quick Download -- now instead of right clicking, hitting Download, waiting for file dialog to pop up and hitting Save, you right click, hit Quick Download, and it's done for you.
Links bar, similar to Mozilla's Page Info -> Links tab. It's a bit primitive at the moment, but it's nice to see they're working on stuff like this.
Fast Forward -- fancy <link rel="next"> UI gadget -- if a site uses said links, the Forward button is turned into a Next button, which is nice for browsing things like search results pages and blogs.
Improved skins support -- auto-install for new skins, more flexible for users (no more.ini editing if you want to rearrange your buttons, for instance), etc. Someone badly needs to Opera 7-ize Minimalist, though, I'm not a fan of the Aqua look, or the bare-bones "Windows" skin that ships with this beta.
The bookmark manager is back, and looking nicer than the Opera 6 one.
I'm quite impressed with this second beta. With betas like this, IE7 better be damn good to not get yawned at:)
Microsoft is part of the W3C, and help make many of these standards. If you look at the acknowledgments you'll see Microsoft is actually a member of the working group responsible for these guidelines.
The ISON command was implemented to provide a quick and efficient means to get a response about whether a given nickname was currently on IRC....
PRIVMSG is used to send private messages between users. <receiver> is the nickname of the receiver of the message.
In rfc1459, dated May 1993, after 4 years of development. Although it is described as a "teleconferencing system", it does sound like it'd match:
"The claim is it's a system where you have a network; you have a way to monitor who's on the network; and if you want to talk to them you hook them up,"
Mozilla- very good, feature packed, but no way near as good as IE - it pains me to say it, but everyone harps on about the fact that IE doesnt support all the standards that Mozilla does. But if people dont seem to be coding in those standards, and are using the broken Microsoft ones, then what the hell is the point?
Vicious circle: IE doesn't support [x], web developers don't use [x] because 90% of users won't see it. Hence IE doesn't need to support [x]. Way to hold up the development of the web by barely supporting 6 year old standards.
Funnily enough they do the same with Outlook and their MIME support.
IE belts along at a pace I have yet to see Mozilla keep up with
In my experience, Gecko is significantly faster than Trident. Phoenix even goes a good way to making the rest of the UI similarly speedy.
and it renders every website correctly, regardless of whether its using an incorrect standard to do it or what.
Erm. Y' know, if we didn't spend so much of our time working around IE bugs, half the sites out there using CSS would probably be unusable with stylesheets enabled in IE. Of *course* it renders websites in pretty much the way we intend -- we spend ages working around the broken box model of IE5 and the broken positioning model of IE6 and the stupid clipping bugs it's covered with and the poor selector support and... you get the idea.
That's not to say the other browsers are pefect, but IE really takes the cake for destructive and annoying to work around bugs.
Re:Things dont kill people..
on
Google vs. Evil
·
· Score: 1
Bombs don't kill people, people make bombs kill people. Ergo I get to keep my own stockpile of nuclear weapons, right?
Oh, well, just clone one of the width/height's at the top and put said measurements in. The [width=".."][height=".."] go by what's specified in the HTML, so there's no conversion needed.
I do actually block 346*280, but that's done specifically for SlashDot's middle-of-the-article ads, rather than every image or iframe, since it's using a table with width/height rather than an img, and I wasn't sure how standard that size was.
I'll see about generalizing it a bit more -- I might change all those iframe/a img's to *, so I catch everything with those dimensions.
Can you elaborate a little on this CSS and how it might be generalized?
How do you mean, generalized? Generalized into what?
The CSS should be pretty self documenting if you know a little of CSS 2 selector syntax.
The parameters you use don't correspond to the IMU values reported on the AIB article, so I assume there is some mapping to be done between IMU and pixels?
I'm not sure what you mean -- all the width/height blocks in there include the new IMU recommendations, except for 180*150 which I just added.
Why do you need to print those books? Read them on your monitor and they are old already in a few months time anyway.
Reading books on screen is typically much less pleasent than reading in dead tree form.
I have tonnes of eBooks, including many from O'Reilly -- I haven't bothered reading them, because it's simply not a nice way to read large quantities of text.
Er, if it's SVCD it's 480 x 576 PAL or 480 x 480 NTSC. Otherwise it's not an SVCD, it's just a random MPEG-2 stream burnt with the SVCD format.
A good SVCD will blow up just fine, especially on a TV.
Don't mistake TeleSync VCD's for DVDrip SVCD's. Just like there's plenty of crappy, low resolution, poorly encoded DivX's, the same can be said for SVCD.
dude, your site doesn't render at all for me (moz1.0rc2/linux)
Fine in Moz 1.0.1 and 1.2.1/Win32 here. Although it does bring up one border rendering regression in 1.2. Also, something's interfering with keyboard input (like CTRL-w)
Upgrade. I don't support prereleases;)
I serve as application/xhtml+xml for Mozilla -- maybe rc2 is buggy with it.
Slashdot Mirror
IE6, Opera 5+, Netscape 4, lynx and w3m also support gzip and compress encoding. w3m even claims to support bzip2 encoding, although that's probably a bit heavyweight for this sort of thing :)
:)
Anyway, gzipping content can easily make for an 8x size reduction on a large page, especially if there's a lot of repetition in there, e.g. lots of tables. Whether this translates to a significant speedup in browsing speed depends heavily on the size of the page and the speed of the connection; certainly on a modem, going from 80k to 10k is very noticable
Actually it struck me as being quite logical, especially for such an important syscall. It means you can get the options with a single pretty much atomic and very cheap call to lstat() -- otherwise you'd need at least an open()/read()/close(), and then you run into worrying about locking while the config file's open.
:)
It did make me do a bit of a double-take when I first saw it though
boxes(1) will draw these for you. par(1) will (try to) fix broken ones.
Actually, malloc.conf is supposed to be a symlink to a file*name* containing options, not a regular file containing them, i.e:man malloc for more flags, and alternative ways to specify them.
No, it's a Barracuda V, not IV.
StorageReview, uh, reviewed the PATA model a couple of months ago; they found it to be slightly quieter than the Barracuda IV, with largely similar performance.
Are IP's really so scarce that companies are resorting to stealing them to stay afloat?!
/32, but just you wait until we have our first nuclear exchanges over those last few /8's!
It's obviously time for us to migrate to IPv6, before this escalates and we have our first full scale war over netblocks.
Oh, sure, it's going to start with this sort of scuffle over a
Logout will require an extra request either way, since logout should be mainly server side -- you don't want to ask the UA nicely if it'll please delete your session cookie or whatever, because it's well within it's rights to tell you to go jump in a lake, and may even pretend it's done so without actually doing anything. Clients can not be trusted.
I hate the way modal dialogs are implemented in most UA's, though. They have no place in a browsing environment -- I want to continue browsing elsewhere and an always on top dialog that demands input and which steals focus SUCKS. Better would be to use JS+DOM/CSS to place an item on the page (maybe replacing the form) to do the confirmation.
Less compatible, but you can probably fairly easily get 90% of clients, with the other 10% getting a traditional extra request.
If the description is small, the title attribute is better. You can even use some DHTML to make it into a pseudopopup on click. Personally I'd just do a normal link with a glossary and let the user hit Back.
Probably the worst use. If they're articles and pictures, they're real content and I'd rather get them in the current page rather than screwing up my normal browsing reflexes and giving me a teeny browser window that may or may not even fit in with my environment.
This is a user interface issue with the UA, nothing to do with HTTP authentication. There's nothing stopping browser developers adding a Logout button to the navbar when it's using HTTP auth.
There's plenty of reason, some of it you also get with frames -- they break the UI the user is used to by breaking Back, maybe disabling parts of the UI to make the window look "cleaner", etc.
Plus they don't work everywhere, so you're already breaking the "access anywhere" concept.
I've been doing this, although I've not advertised anywhere.
Build and set up a machine -- £50 + parts.
Troubleshoot some problem -- £15-£30+, depending on complexity.
The nice thing about this is you'll tend to get more clients the more you do, as people go around telling others about how you fixed their computer and how they should give you a call whenever they have problems.
Yup, you're right. It seems to look for <link rel="next">, <a rel="next">, and <a>next</a>.
Yes. Microsoft's web team leaves quite a lot to be desired. But then, so do most other so called "professional" web developers who wouldn't know a standard if it jumped up and down screaming "I'm a standard! I'm a standard!".
Tantek's website is awful. It does at least degrade well when I turn off CSS, but it's not very nice to use. He even has a splash page ffs
But Tantek is Microsoft. So is Tim Lacy, and Chris Wilson, and Ed Tecot, and Laurie Anna Kaplan, and David Meltzer, and Stephen Waters, and Scott Isaacs, and.. well, you get the point. Microsoft are not a single entity out to be incompetent and ignorant, it's a whole bunch of people, mostly incompetent and ignorant. Just like any other random collection of 50,000 people
I'm quite impressed with this second beta. With betas like this, IE7 better be damn good to not get yawned at
Microsoft is part of the W3C, and help make many of these standards. If you look at the acknowledgments you'll see Microsoft is actually a member of the working group responsible for these guidelines.
The patent was filed in 1997.
In rfc1459, dated May 1993, after 4 years of development. Although it is described as a "teleconferencing system", it does sound like it'd match:
Vicious circle: IE doesn't support [x], web developers don't use [x] because 90% of users won't see it. Hence IE doesn't need to support [x]. Way to hold up the development of the web by barely supporting 6 year old standards.
Funnily enough they do the same with Outlook and their MIME support.
In my experience, Gecko is significantly faster than Trident. Phoenix even goes a good way to making the rest of the UI similarly speedy.
Erm. Y' know, if we didn't spend so much of our time working around IE bugs, half the sites out there using CSS would probably be unusable with stylesheets enabled in IE. Of *course* it renders websites in pretty much the way we intend -- we spend ages working around the broken box model of IE5 and the broken positioning model of IE6 and the stupid clipping bugs it's covered with and the poor selector support and... you get the idea.
That's not to say the other browsers are pefect, but IE really takes the cake for destructive and annoying to work around bugs.
Bombs don't kill people, people make bombs kill people. Ergo I get to keep my own stockpile of nuclear weapons, right?
Oh, well, just clone one of the width/height's at the top and put said measurements in. The [width=".."][height=".."] go by what's specified in the HTML, so there's no conversion needed.
I do actually block 346*280, but that's done specifically for SlashDot's middle-of-the-article ads, rather than every image or iframe, since it's using a table with width/height rather than an img, and I wasn't sure how standard that size was.
I'll see about generalizing it a bit more -- I might change all those iframe/a img's to *, so I catch everything with those dimensions.
How do you mean, generalized? Generalized into what?
The CSS should be pretty self documenting if you know a little of CSS 2 selector syntax.
I'm not sure what you mean -- all the width/height blocks in there include the new IMU recommendations, except for 180*150 which I just added.
In your userContent.css in Moz/Opera:
table table table[width="346"][height="280"] {
display: none !important;
}
Hint to advertisers: I don't bother blocking small adverts, and I might actually *read* textads.
Reading books on screen is typically much less pleasent than reading in dead tree form.
I have tonnes of eBooks, including many from O'Reilly -- I haven't bothered reading them, because it's simply not a nice way to read large quantities of text.
Also try newzbin's new system, which is rather like a scaled up http://alt.binaries.nl/
Er, if it's SVCD it's 480 x 576 PAL or 480 x 480 NTSC. Otherwise it's not an SVCD, it's just a random MPEG-2 stream burnt with the SVCD format.
A good SVCD will blow up just fine, especially on a TV.
Don't mistake TeleSync VCD's for DVDrip SVCD's. Just like there's plenty of crappy, low resolution, poorly encoded DivX's, the same can be said for SVCD.