just got a wild hair - is it possible to group this way?

[sfullman]

suppose you have a db where records are being added sequentially. you have a timestamp field recording the time of entry. you also have a field that could have a range of values, say 1 to 5.

That value might be 1 for a while, maybe 100 records, then it might change to 2 for a while, then to 5, then back to 1. Let us say that the changes are totally arbitrary and the number of records in any change might be 1, might be infinite.

Is it possible to group by these changes? I have an application where this would be useful. Note we're not grouping by the value of the field but by each time in the time sequence the value distinctly changes.

This is what happens when you're up at 4 in the morning..

[bradgrafelman]

What do you mean "group by" ? It sounds like you're talking about modifying how the data is displayed rather than the data itself. In that case, do the logic to detect a change in the PHP, e.g.:

$lastNum = null;
while($row = mysql_fetch_assoc($query)) {
    if($row['num'] != $lastNum) {
        // output some boundary/separator/etc. to denote change

    $lastNum = $row['num'];
}

// do whatever you want with the data
}

[sfullman]

actually no I'm wanting to group by a change in data - i.e. run group-related functions on data which is grouped by a chronological change in a field's value, so that, if there are 5 records contiguously where fieldA = 1 then those are grouped, but if fieldA changes to = 2 (in chronological order), those records are grouped, etc.

And I'd like to do it without a secondary PHP operation

It just seems so elegantly intuitive conceptually that it wouldn't suprise me if mySQL could do it.

Sam