From: Lee j. <jun...@gm...> - 2008-02-01 11:33:39
|
Hi, I modified some code related with time-zone (modified time() to gmtime()) and committed it at sf.net. Now, It looks like run well. (Number of changesets (Day) and Number of changesets (Month)) But there are another miss calculated statistics. I have other kernel git-tree for gitstat testing. I made a graph and statistics for Top contributors (Kernel release v2.6.24) The results are... 56 David Woodhouse 41 (tree.celinuxforum.org) 121 David Woodhouse 20 (my own gitstat) 16 Alan Cox 91 (tree.celinuxforum.org) 14 Alan Cox 104 (my own gitstat) Of course, I am making up-to-date both gitstat. Our first mission for gitstat is that it should give accuracy statistics. So, We should focus on that more than other issue. Thanks, -----Original Message----- From: Paul Vriens [mailto:pau...@gm...] Sent: Friday, February 01, 2008 4:29 PM To: ??? Subject: Re: [Gitstat-devel] [gitstat-devel]Timezone problem Paul Vriens wrote: > Lee jungseung wrote: >> Hi, >> >> The gitstat use 'mktime()' for making time information. >> e.g. $cart_time_start=mktime(0,0,0,1,1,$_GET['chart_parameter2_year']); >> As you know, the $cart_time_start is something like '2008.1.1 '2008.12.31 >> (a kind of time standard?) >> >> According to server setting, Time zones are different. So, >> gitstat(using same git-tree) shows different statistics. >> >> In my opinion, We need to use same time standard. >> >> In case of using mktime(), How about to use gmmktime()? >> http://kr2.php.net/gmmktime >> With gmmmktime(), we could maintain some standard for making statistics. >> How about your opinion? >> >> Thanks, > > Hi Lee, > > I see you've started committing some of those timezone changes. You > accidentally reverted a change in index.php (moving stuff over to > libgather.php). Can you fix that? > > Cheers, > > Paul. > Hi, I also see that the current index.php (and chart.php) doesn't show the correct graph anymore for the 'commits per month'. When I go to http://tree.celinuxforum.org/gitstat/ I see that this graph end for Dec/07 where it should have been Jan/08. This wasn't the case before. Cheers, Paul. |