Steven Noels wrote:
> Steven Noels wrote:
>
>> Jeff Turner wrote:
>>
>>> We really need stats on 4.x usage for xml.apache.org to make a good
>>> judgement.
>>
>>
>>
>> I'll try and make a browser usage analysis for the month of
>> september. I would be using analog or webalizer - does anyone else
>> has killer tips for log analyzers?
>>
>> </Steven>
>
>
> I put up an analysed log of 1 day (3.868.269 requests, 271.678
> gigabytes) on http://www.apache.org/~stevenn/browserstats/
>
> I'm not a browser User Agent dictionary, it seems to me that the first
> non-IE, probably-dated browser is identified by "Mozilla/3.01
> (compatible;)", totaling 2.11% of that day. And Mozilla at 1.26% :-(
Hi Steven,
As you see from the pie chart... It's quite useless to just count unique
browser agents. Without groupings, these statistics do not mean much.
I think all these agents should be grouped into several larger categories:
* MSIE 6.0+
* MSIE 5.5+
* MSIE 5.0+
* MSIE 4.0+
* Older MSIE
Similar groupings can be done for Netscape browsers, and for mozilla
ones (which has lots of different unique user-agents strings).
Then, pie chart will look different, a lot.
What do you think? Do you need some help to work this out?
Vadim
> OK, that's one day, and due to the sheer size of these logfiles and
> the current load-issues of daedalus I'm reluctant to run it across an
> entire month's worth of logs, but this already learns us something.
>
> The '10%' of older browsers on apache.org has been an urban city
> legend, IMO.
>
> </Steven>
|