Hi, Rene I took a quick look over your code ... I kind-of like the idea having all logging at one place, but the code is a bit too messy if you ask me :) If you would have put it on github and I would fork it - the first thing I'd do is trying to get rid of the hm-lib and rewriting all in a bit more object-oriented style, but that's just personal taste ;) Specially the function rajmvServiceLog_graphs_raphael_calculateData() with a code of ca. 280 lines is quite long ... Additionally I think that setting an error-handler who's just returning false is not a good way. Why not disable error-handling or write code that produces no errors? I think it would be good to mention that you're using the library adodb ( http://adodb.sourceforge.net/) and sitewide_rv (is it that? http://mediabeez.ws/) or am I guessing wrong? You're talking about a sql-file ... has my anti-virus-program removed something here? Please don't see that as destructive critic, but as hints what I would do if I get to do with this code. Bye Simon On Mon, May 21, 2012 at 12:46 PM, rene7705 <rene7705@xxxxxxxxx> wrote: > Hi. > > I wasn't happy with the fact that Google Analytics doesn't record many > of the hits on my sites, so I decided to roll my own analytics > software, and opensource it. > It's now in a state where I can request early comments. > You can view a demo at http://mediabeez.ws/stats (under construction, > may fail at times) (browser compatibility may be an issue, I built it > with chrome, should work in firefox too, but won't (ever) in IE.) > If you click on the graph while the details for a given day are > visible, you will see the errors for that day in a DIV below the > graph. > > Normally you'd hide the error details and $hits for anyone's who's not > registred as a developer, of course. I've turned it on for all at the > moment, so you can comment on that feature and review my results array > $hits. > > I've opted, for simplicity of design, to store all settings I thought > could be remotely interesting in a mysql db (see attached sql init > file) (which is filled from PHP every time a page is delivered, and > again from JS when the page has initialized fully, or does a HTML5 > History API location.href change), and then use PHP to retrieve all > rows for a given datetime-range, and do the totals calculations in a > php loop. > I'm only keeping 1 row from the db in memory at any given time, but > I'm building up a large deep array with the totals information in the > php loop that goes over the rows. > I'm wondering if this is a good approach, though. Maybe I should let > the totals be calculated by the mysql server instead. > > I was thinking to let the totals calculations stay in php, and be > executed from a cron job every hour. Only for the current month would > you need to re-calculate every hour. > I haven't figured out yet how to for instance only re-calculate the > last hour, store results per hour, and then calculate the day and > month totals from that "hourly cache" data when needed. > I don't have a clue about how big companies do their totals > calculations (for sites that get way more hits than mine, which is > something i'd like to be able to support with my own analytics code), > and would like to know how big companies do this. > > I've included the relevant draft code as attachment files to this > mail, for your review. Please let me know if I have forgotten to > include relevant code.. > > As always, thanks for your time, > Rene > > -- > PHP General Mailing List (http://www.php.net/) > To unsubscribe, visit: http://www.php.net/unsub.php >
