← All modules

The POSIX password-database calls, reshaping the platform's struct passwd into the field order Perl's getpw built-ins use.

Version

0.0.11 zef:lizmat

Depends

none beyond the core

License

Artistic-2.0

Its own test suite

2 files, green

Checked

2026-09-15 against Raku++ 3.28.0 and Rakudo 2026.08

Where it lives

raku.land · source

Install it #

$ rakupp install P5getpwnam

zef install P5getpwnam writes the same store; either installer leaves the module usable by both engines.

What it is for #

Resolving a user name to a uid, or a uid to a home directory, means reading the password database — and struct passwd has a different layout on BSD, on Linux and on generic Unix, so a NativeCall binding has to pick one at load time.

This distribution does that, and presents the result in Perl's field order.

Looking a user up #

use P5getpwnam;

# uid 0 exists on every POSIX machine. Nothing here prints a name,
# a home directory or a shell.
my @p = getpwuid(0);
say 'fields            : ', @p.elems;
say 'name is defined   : ', @p[0].defined;
say 'uid               : ', @p[2];
say 'gid is an Int     : ', @p[3] ~~ Int;
say 'home looks absolute : ', @p[7].starts-with('/');
say 'shell is defined  : ', @p[8].defined;
say '';
say 'scalar by uid gives the name : ', getpwuid(Scalar, 0) eqv @p[0];
say 'scalar by name gives the uid : ', getpwnam(Scalar, @p[0]);

fields            : 10
name is defined   : True
uid               : 0
gid is an Int     : True
home looks absolute : True
shell is defined  : True

scalar by uid gives the name : True
scalar by name gives the uid : 0

getlogin() is bound too, as a direct native call.

Enumerating #

use P5getpwnam;

say 'setpwent : ', setpwent();
my $n = 0;
loop { my @e = getpwent() or last; $n++ }
say 'endpwent : ', endpwent();
say 'entries walked : ', $n > 0;

setpwent : 1
endpwent : 1
entries walked : True

The same empty-list sentinel as the rest of the P5get* family: a failed lookup and the end of the enumeration are indistinguishable.

The one thing to know #

The record's arity is not portable, and two of its fields are hard-coded fillers.

use P5getpwnam;

my @p = getpwuid(0);
say 'this kernel : ', $*KERNEL.name;
say 'record length here : ', @p.elems;
say '';
say 'on darwin and freebsd the list is 10 long:';
say '  ($name, $passwd, $uid, $gid, $quota, $comment, $gecos, $dir, $shell, $expire)';
say 'on linux and generic unix it is 9 — the trailing $expire is absent.';
say '';
say 'so the same destructuring assignment binds different things per platform.';
say '';
say 'and fields 4 and 5 are literals the module writes in:';
say '  field 4 (quota)   is always : ', @p[4].raku;
say '  field 5 (comment) is always : ', @p[5].raku;
say '  the real GECOS text is field 6, one later than a Perl one-liner';
say '  that says "the fifth element is the comment" would reach for.';

this kernel : darwin
record length here : 10

on darwin and freebsd the list is 10 long:
  ($name, $passwd, $uid, $gid, $quota, $comment, $gecos, $dir, $shell, $expire)
on linux and generic unix it is 9 — the trailing $expire is absent.

so the same destructuring assignment binds different things per platform.

and fields 4 and 5 are literals the module writes in:
  field 4 (quota)   is always : 0
  field 5 (comment) is always : ""
  the real GECOS text is field 6, one later than a Perl one-liner
  that says "the fifth element is the comment" would reach for.

Walking the whole database finds no entry anywhere with a non-zero field 4 or a non-empty field 5 — they are not read from the operating system at all.

Never write getpwuid(0).elems into a test, and index the GECOS field as 6.

Where the two engines differ #

Nowhere. Every lookup, both scalar forms, the enumeration and the field analysis produced identical output on Raku++ and Rakudo.

Two things to keep out of any example that has to run anywhere: uid 4294967294 is not a safe "absent uid" — it is nobody on macOS — and the record arity, as above. getlogin() returned a defined Str even with no terminal attached, which is not something to rely on.