__DATA__ Inconsistency

bichonfrise74 has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: __DATA__ Inconsistency by moritz (Cardinal) on Oct 02, 2008 at 23:24 UTC
You can write this as `my $test = do { local $/; <DATA> }; print $test;` [download] See $/ for an explanation of what it does.	[reply] [d/l]
Re: __DATA__ Inconsistency by GrandFather (Saint) on Oct 02, 2008 at 23:30 UTC
I'd add in `use warnings;` as a matter of course, but that won't make any difference to your problem. I'd `$test .= $_ while <DATA>;`, but that won't make any difference either (although it means you don't need to init $test in the previous line). Rather than the while loop I'd more likely write: `my $test = join '', <DATA>;`, but that won't make any difference either (Update: actually if I were a couple more cups of coffee through the day I'd use $/ as moritz suggests). In fact, I can't see how the code you've shown could fail in the manner you describe. Does your sample code actually fail, or have you golfed it down from a larger piece of code that does fail? If the sample is golfed, try generating a sample that actually does fail and show us that. Perl reduces RSI - it saves typing	[reply] [d/l] [select]
Re^2: __DATA__ Inconsistency by bichonfrise74 (Vicar) on Oct 02, 2008 at 23:44 UTC
It does fail and it would add in the $_ some messages that occurred early on, but again it is so inconsistent. I wish I could re-produce it. What I did was I "initialized" $_ to be safe. `$test = qq(); $_ = qq(); $test = $test . $_ while(<DATA>);` [download]	[reply] [d/l]
Re^3: __DATA__ Inconsistency by GrandFather (Saint) on Oct 02, 2008 at 23:51 UTC
In your sample code there is no "early on". Show us the actual code that is failing rather than what you consider to be a representative snippet. The code you have shown can not fail in the way you describe! Perl reduces RSI - it saves typing	[reply]
Re: __DATA__ Inconsistency by cdarke (Prior) on Oct 03, 2008 at 04:59 UTC
The only way I can get previous values of $_ printed is to slightly modify the loop: `use warnings; use strict; my $test = qq(); $_ = 'xxxx'; do{$test = $test . $_ } while (<DATA>); print "$test"; __DATA__ Hello there. This is me.` [download] Are you perhaps doing that?	[reply] [d/l]
Re^2: __DATA__ Inconsistency by bichonfrise74 (Vicar) on Oct 04, 2008 at 03:41 UTC
It is very possible that I'm doing what you've just written. But the code is too long for me to post here. Anyway, thanks for all your inputs. I really appreciate it.	[reply]
Re^3: __DATA__ Inconsistency by ikegami (Patriarch) on Oct 04, 2008 at 04:07 UTC
`do BLOCK while EXPR;` [download] is uniquely special case of `EXPR while EXPR;` [download] The `do` variant is a bottom tested loop. The loop body will always execute at least once, and will do so before executing the condition expression.	[reply] [d/l] [select]
Re: __DATA__ Inconsistency by tmaly (Monk) on Oct 03, 2008 at 04:53 UTC
try using tell and seek to debug issues with __DATA__ I ran into problems where I was reading from __DATA__ in a module multiple times and one the second read the position had moved to the bottom. However this may not be your case, but using tell to get the position and printing it out might shed some light.	[reply]
Re: __DATA__ Inconsistency by massa (Hermit) on Oct 03, 2008 at 02:11 UTC
What you are calling "old values" are IMHO the "current values". Your first line came blank because, well, you put a blank line as the first line of `__DATA__`. delete the line between `__DATA__` and `Hello there.` and all will work as you want :-) []s, HTH, Massa (κς,πμ,πλ)	[reply] [d/l] [select]
Re^2: __DATA__ Inconsistency by GrandFather (Saint) on Oct 03, 2008 at 02:21 UTC
You are describing a "fix" to a deterministic behavior which you inferred to be the issue the OP is talking about. However the OP clearly states that his problem is non-deterministic - your fix is for the wrong issue. Perl reduces RSI - it saves typing	[reply]