Re^2: Closure producing 'Modification of read-only value' error

No, don't do that.

Never local $_ in a subroutine. If you want to use the magic assignment you should do local *_. Of course you have to unpack anything you want from @_ first.

If you get into the habit of doing local $_ inside you subroutines it will seem to work fine - maybe for years. Then one day someone will call one of your subroutines inside a for() loop that is iterating over the values of a tied array or hash. Very odd things will start happening in their program.

Comment on Re^2: Closure producing 'Modification of read-only value' error Select or Download Code

Replies are listed 'Best First'.
Re^3: Closure producing 'Modification of read-only value' error by tlm (Prior) on Jun 11, 2005 at 16:39 UTC
What about the implicit localizations of `$_` done by `map`, `grep`, `for`, etc.? Do they do The Right Thing®? I think it is surprising (hence bad) that the `while ( <FH> )` idiom doesn't do an implicit localization of `$_` like the looping constructs listed above. Is there a fundamental reason for this, or is it due to a severe tuit shortage? the lowliest monk	[reply]
Re^4: Closure producing 'Modification of read-only value' error by nobull (Friar) on Jun 11, 2005 at 19:09 UTC
`map`, `grep`, `for` etc do the right thing. The magic while construct does not make `$_` and alias - it assigns to it. This is fundamentally different. That said, it bugs me that it doesn't do implicit localization.	[reply] [d/l] [select]
Re^3: Closure producing 'Modification of read-only value' error by kaif (Friar) on Jun 13, 2005 at 02:03 UTC
I'm sorry, maybe I'm dense, but what exactly is different when you iterate over a tied array or hash. What are these "very odd things" and why won't a simple `local $_` do? This is very interesting to me.	[reply] [d/l]