A Standard for Greatly Reducing HTTP Connections

Note that you don't have to run separate physical servers for the dynamic vs static content. It's as simple as setting up Lighttpd on the server and having it respond to requests on port 8080 and then linking to your images as "http://ozzu.com:8080/img/logo.png" for example. You don't even have to touch Apache's configuration, which means you can give it a test run with minimal effort or risk.



I feel you there. At one point one of the sites we're responsible for was at the point that every single additional byte was translating into an additional ~25MB of bandwidth for the month because we couldn't use gzip or cache a lot of files. The traffic was insane. Apache would have crippled under the load. Lighttpd let us run with a load average of around 2.5 for a 4 'core' server. We so much as went in and switched HTML entities to their UTF-8 strings where possible to conserve bandwidth, though. It was a nightmare.

If you've already got the truck and a warehouse (browser cache), wouldn't it make sense to fit that truck to carry eggs in bulk and maybe grab some milk in the process ?



You'll have to enlighten me about that. It sounds like something to do with email and I don't do much with email.

Noticed later post while reading.



You'll have to tell us about that some time.

Writing out



uses up 5 more bytes than...





In order to save bandwidth we were stripping as much extra weight off of the page as possible.

If you properly encode the page, you can do without many special character entities. You still need to escape things like ampersands with &, but you don't need to do it for every special character.

@effim: (Completely off topic)... on your site, under "User-Interface Kung-Foo" you got the following text...

I'm sure that the bolder word was supposed to be translate

@mods: Sorry to go off topic...




Makes sense... you said UTF-8?

Thanks for the misspelling heads up. There's never enough time to do our own site properly.

I prefer using UTF-8, yes. It's by far the most versatile character encoding for the web.

alright.

Since I have never encoded a file to anything... I wouldn't know how to do that... want to share the wealth?

That doesn't seem to do that... when I try • I get either � or ? or simply a square...

be sure to save the source text file on the webserver to utf-8 too, otherwise the • character will be corrupted itself, eg. in notepad use the dropdown titled "save as type" in the save as dialogue

Ahh ok. I was thinking "Ӓ" instead of "&named;" for some odd reason. I remember seeing someone hell bent against   saying it was a conspiracy by the web hosting companies years ago now that you mention that.

LMAO. I suppose if network providers billed by the byte (like I'm sure AT&T would have loved), I could have seen that becoming a pretty significant argument. Of course, in that world, Google wouldn't even be feasible, let alone YouTube.

This isn't really the appropriate place for this discussion. Open up another thread and I'd be happy to help, as would plenty of other folks around here, I'm sure.

Simply put, character encoding allows for us to ensure that regardless of the character set used (like Croatian characters or the Roman alphabet), the proper character displays on the screen. It doesn't change the actual ascii data that is saved in any way, it just sets a flag for whatever software is reading it so that it can properly interpret the ascii characters and display the appropriate symbol/letter/number. Yes, that's simply put. It's a pretty hairy issue.

Did