sort...then rsort/arsort confusing results

Joseph_Sliker

Hi; Here's the deal. I have sets of dated observations in my database. I will have a variable count of observations per person over time. (ie. starting observation date, and then one or more subsequent observations).

When I do my mysql query to get data, I deposit all the dates (in unix_timestamp form) in an array (called, say; $mydates). I want to sort($mydates) and then use $mydates[0] as $start_date=$mydates[0];

print_r($mydates) looks like this: "Array ( [0] => 1347858000 [1] => 1371013200 )"

so far so good.... and in this case $mydates[0] shows up in calculations as 1347858000 like it should

BUT I then want to arsort($mydates) (ie. $myRdates=arsort($mydates)) so that I can use $end_date=$myRdates[0] as the final date for the series (ie to calculate number of days over which observations took place).

print_r($myRdates) looks like this: "1"

I don't get what's going on here...

bradgrafelman

Joseph Sliker;11029997 wrote:
I want to sort($mydates) and then use $mydates[0] as $start_date=$mydates[0];

Any reason why you didn't just let your DBMS sort the results for you?

Joseph Sliker;11029997 wrote:
BUT I then want to arsort($mydates) (ie. $myRdates=arsort($mydates)) so that I can use $end_date=$myRdates[0] as the final date for the series

Er... why? If you've just sorted the array in ascending order, why not just look at the last element in the array?

Joseph Sliker;11029997 wrote:
(ie to calculate number of days over which observations took place).

Again, a DBMS like MySQL is perfectly capable if telling you the difference between two timestamps in days, so... why not let your DBMS do all the work for you?

Joseph_Sliker

I'm not very good with complex queries, so I try to do this in a way that gets me a whole mess of information via one fairly simple query that runs fairly quickly involving lots of groups, several kids in each group, and then multiple scores per kid on multiple dates. The problem is that the reporters are inconsistent and the data is spotty, so I throw the resulting data into several multidimensional session arrays to group and display the data in a variety of ways. (this is helpful in pointing out whose data is tardy, missing or otherwise incorrectly reported).

You are right though: I figured out a work-around that gets me (perhaps more efficiently to the last element...while the data is cycling into the $mydates[] array elements, so long as the source is correctly ordered, I can keep overwriting a variable called $end_date, which, by the end of the cycle should be my final/most recent observation date:

$xdates=array_values($_SESSION[testdates_array][$plan_id][$kid]);
asort($xdates); //this should put all the unix_timestamp strings in the correct order
foreach($xdates as $tdate) {
$mydates[]=$tdate; // earliest observation should be $mydates[0]
$post_date=$tdate; // finds the last date
}
$calcdiff=round(($post_date-$mydates[0])/86400); //ta-dah!

bradgrafelman

Note that you could just use [man]end/man to get the last element in the array rather than the (seemingly) pointless loop.

Also bear in mind that not every day is 86,400 seconds in length. 😉

Joseph_Sliker

so simply (end($xdates)-$xdates[0]) would give me the time span I'm looking to convert to days? Hmmm... gotta try this.