csh_glob#

Expand a pattern using the default flag set — the engine behind Perl’s built-in glob.

Synopsis#

use File::Glob qw(csh_glob);
my @files = csh_glob('*.pl');

What you get back#

A list of matching filenames. Flags always come from $File::Glob::DEFAULT_FLAGS (initially GLOB_CSH); csh_glob takes no flag argument. Sort order, brace expansion, tilde expansion, and the rest follow whatever DEFAULT_FLAGS currently holds.

You almost certainly want bsd_glob or the angle-bracket operator <...> instead. csh_glob exists because Perl’s built-in glob dispatches through it; calling it directly is documented as “don’t unless you know what you’re doing.”

Examples#

use File::Glob qw(csh_glob);
my @matches = csh_glob('*.txt');        # uses DEFAULT_FLAGS

## Flip case sensitivity globally, then call

use File::Glob qw(csh_glob :nocase);
my @any = csh_glob('ReadMe*');          # matches README, readme, ReadMe, ...

Edge cases#

Called with no arguments: returns the empty list.
Unlike upstream, does not split the pattern on whitespace into multiple sub-patterns; the whole string is treated as one pattern.
Does not maintain per-op iterator state, so it cannot be driven one-result-at-a-time in scalar context.

Differences from upstream#

No whitespace-splitting of patterns. Upstream csh_glob tokenises "a* b*" into two patterns; pperl treats it as a single literal pattern. Use bsd_glob('{a*,b*}') if you need multiple patterns in one call.
No scalar-context iterator. The per-op result cache that lets upstream’s while (my $f = <*.c>) yield one name per iteration is not implemented — scalar context returns the match count. Covered by t/81-xs-native/File/Glob/*.t.