You need to load in a module that might not be present on your system. This normally results in a fatal exception. You want to detect and trap these failures.
Wrap the require
or use
in an eval
, and wrap the eval
in a BEGIN block:
# no import BEGIN { unless (eval "require $mod") { warn "couldn't load $mod: $@"; } } # imports into current package BEGIN { unless (eval "use $mod") { warn "couldn't load $mod: $@"; } }
You usually want a program to fail if it tries to load a module that is missing or doesn't compile. Sometimes, though, you'd like to recover from that error, perhaps trying an alternative module instead. As with any other exception, you insulate yourself from compilation errors with an eval
.
You don't want to use eval
{
BLOCK
}
, because this only traps run-time exceptions and use
is a compile-time event. Instead, you must use eval
"string"
, to catch compile-time problems as well. Remember, require
on a bareword has a slightly different meaning than require
on a variable. It adds a ".pm"
and translates double-colons into your operating system's path separators, canonically /
(as in URLs), but sometimes \
, :
, or even .
on some systems.
If you need to try several modules in succession, stopping at the first one that works, you could do something like this:
BEGIN {
my($found, @DBs, $mod);
$found = 0;
@DBs = qw(Giant::Eenie Giant::Meanie Mouse::Mynie Moe);
for $mod (@DBs) {
if (eval "require $mod") {
$mod->import
(); # if needed
$found = 1;
last;
}
}
die "None of @DBs loaded" unless $found;
}
We wrap the eval
in a BEGIN block to ensure the module-loading happens at compile time instead of run time.
The eval
, die
, use
, and require
functions in Chapter 3 of Programming Perl and in perlfunc (1); Recipe 10.12; Recipe 12.3