Re^2: Getting the right count

I can see two potential disadvantages to this:

The file is needlessly loaded into memory all at once. Nothing to worry about with the files the author quoted, but why do it if a while() will do just as well?
If there is some kind of format violation like a blank line or one that doesn't start with a number, the variant using split will always spit out a warning while yours will silently use zero (which may be what the condition is checking for).

Comment on Re^2: Getting the right count

Replies are listed 'Best First'.
Re^3: Getting the right count by ww (Archbishop) on Dec 31, 2011 at 12:21 UTC
mbethke: The first (++) is, indeed a "potential" disadvantage... but it is easily remedied if the need arises. In any case, slurping the file is not the point; use of a regex to ID files is. The second -- IMO -- is pretty much specious. Did you try inserting a "format violation" (or several)? The only decimal cases which effect the output that I've discovered are 1) a record which consists solely of a number (clearly, a case which is cause for concern) or 2) a number using a thousands separator. (Admitted: I followed the sample data, but, as you can see from the regex, numbers with 4 or more digits satisfy the test so long as any thousands separator is omitted. And if it's present, the regex needs minor modification, followed by a function to remove the offending punctuation. If the first is a case to worry about, spit out the entire record whenever a number satisfies the test. If the latter, the data is sufficiently suspect that its content should be validated... which is a different kettle of fish. So too is the case in which the numbers are binary or or octal or hex or lakhs or .... But, again, illustrating all that seemed OT to me; TIMTOWTDI is the point.	[reply]
Re^4: Getting the right count by mbethke (Hermit) on Jan 02, 2012 at 18:34 UTC
Sorry for the late reply, I was away a couple of days! Happy new year BTW :) Yeah, I tried inserting both a blank line (counted as zero) and one containing "foo800bar" (counted as 800). Sure, that's just as easily fixed by tweaking the regexp as the file slurping but I'm not sure it would be quite as easy for the OP, especially as he'd have to spot these things first. If the original solution both works for files of arbitrary size and does some simple format validation without any extra effort, IMHO that's the way to go.	[reply]