This site is now 100% read-only, and retired.

XML logo

Code articles?
Posted by Steve on Tue 14 Nov 2006 at 15:52
Tags: none.

In the past I've writen/posted a couple of articles on various "code" related topics. I figure thats fair game, because a lot of administration is automation.

After sending mail to a user today describing how some of the tag cloud is implemented I'm wondering if it would be interesting for people to learn how this site is implemented/extended/managed?

I think not, but I'm not sure. I could probably string something together talking about "intelligent URL design", and performance bottlenecks.

e.g. We were on the front-page of digg last week and handled it just fine. One article read about 16,000 in a day, which didn't cause any problems at all for the rest of the site as far as I can see. (I was away travelling either way.)

Most of this just comes from using memcached, along with bandwidth limiting for Apache.

Maybe I could just mull it over for a few days, who knows I could just use it for the 2 year anniversary, although technically that is well past. I'll just pretend, again, that we were born in December. Worked last year ;)


Comments on this Entry

Re: Code articles?
Posted by dkg (216.254.xx.xx) on Fri 17 Nov 2006 at 17:26
[ View Weblogs ]
I'd be interested in hearing your technical observations, goals, and problems surrounding the creation and modification of this site and yawns. I was introduced to memcached from your article. And i want to hear what you have to say about intelligent URL design, since i think many of us system administrators at some point need to craft various bits of web interface for our users these days.

[ Parent ]

Re: Code articles?
Posted by Steve (62.30.xx.xx) on Sat 18 Nov 2006 at 13:54
[ View Weblogs ]

Intelligent design is probably a bad description, but basically it is about making sure that links are collected in a "nice" fashion, but remembering about spiders.

This is why we have the pattern:


But we break that completely for RSS feeds of weblogs which are referred to as:


The reason? We can use robots.txt to make the good robots not spider them (since generation is intensive) and use technical measures to stop the bad robots.

Similarly most of the "editing" links are collected beneath /edit, (eg. /edit/weblog) so that I can use robots.txt to block spiders from that space.) Nesting things like that works really nicely and it isn't terribly user-visible either.

There are a lot of examples of things like that on this site, where things break for non-obvious reasons. I've put a lot of thought into whether some links belong beneath "users" or elsewhere. For example /view/messages is a URL which isn't tied to a user, although it does require a user account to use.

I hope that explains a little.


[ Parent ]