Re: Seemingly weird regex problem

Tim Boring wrote: > On Thu, 2005-01-20 at 13:41, Jason Wong wrote: >> On Friday 21 January 2005 01:52, Tim Boring wrote: >> >> Well the biggest problem in your code right now is your incomprehensible >> (to >> me anyway) use of the switch construct. For a start I've no idea why >> you're >> using ... >> >> > switch ($line) >> >> ... when your tests do not involve $line >> >> > case ($total_counter <= 5): > > I see your point, but the other tests do involve $line. This one that > you reference I'm using for a "special need"--basically so I leave in > the initial column headings from the report, because they would match > several of the tests that would discard those lines. My current theory is this: PHP takes $line and compares it as a scalar to the return value of your expressions. Thus, something like: ($total_count <= 5) which returns a BOOLEAN is compared to $line, which, with AKRN at the front, in boolean turns into 1. However, when you get to ereg_match, which returns, err?, integer, PHP type-casts to integer, at which point AKRN turns into 0. This is what I meant when I said it was just too damn confusing to mix-n-match a value in the switch ($x) with an expression in the case : I'd say use expressions with switch (TRUE) or swith ($x) with constants but don't mix-n-match unless you really want to confuse yourself. I suppose if *ALL* the variables/expresions are of the same data-type, it would be Not So Bad since there will be none of the PHP auto-type-conversion to muddy the waters... -- Like Music? http://l-i-e.com/artists.htm