2009/2/9 Daniel Brown <danbrown@xxxxxxx>: > On Mon, Feb 9, 2009 at 15:21, Stuart <stuttle@xxxxxxxxx> wrote: >> >> I would advise you against wasting your time because there is no >> reliable way to tell what systems a server is actually using to serve >> pages. Nearly all sites I work on these days use techniques to remove >> extensions from URLs, and a fair few hide all details of the server >> software for a variety of reasons. Since you can't say with any >> certainty what your margin of error is, the numbers will be >> meaningless as a measure of language usage. > > Right, but adjusting the math as appropriate should be fine. If a > site doesn't report either PHP or ASP, for example, don't include it > in the count. If we have 100 sites that we spider and 23 report > having PHP capabilities, 16 report being able to support ASP, fifty > (half) having no response, we know that 46% of the total can serve PHP > code while 32% can serve ASP, because we will only record the total > based upon responses. I'm sure there are flaws in this logic.... > which is why I'm thinking aloud here. ;-P You used the right word there... "can" as in not necessarily is! Your results would indicate capability rather than actual usage and even then you have no way to determine your margin of error. Having said all that I would guess that any numbers that look half-way decent would probably convince the people Tedd is looking to sway. But they still won't be anyway near accurate. >> Having said that it's been a while since I wrote a spider, shame I >> don't have time to have a go at the moment. Have fun ;-) > > It's just going to be proof-of-concept code.... certainly nothing > worth pushing Nielsen out the door. Indeed, but I find coding without QA requirements is a great way to relax. -Stuart -- http://stut.net/ -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
