Beefy Boxes and Bandwidth Generously Provided by pair Networks
Pathologically Eclectic Rubbish Lister
 
PerlMonks  

Re: RFC: Defined-Or for before Perl 5.10

by afoken (Chancellor)
on Oct 30, 2010 at 08:25 UTC ( [id://868433]=note: print w/replies, xml ) Need Help??


in reply to RFC: Defined-Or for before Perl 5.10

I see a little problem with using a function instead of an operator: Boolean shortcuts won't work, the function call will always evaluate both sides.

my $pid=fork() // die("Can't fork: $!"); # die()s when fork() failed

vs.

my $pid=defined_or(fork(),die("Can't fork: $!")); # always die()s, twice when fork() was successfully evaluated before d +ie().

Alexander

--
Today I will gladly share my knowledge and experience, for there are no sweeter words than "I told you so". ;-)

Replies are listed 'Best First'.
Re^2: RFC: Defined-Or for before Perl 5.10
by molecules (Monk) on Nov 01, 2010 at 13:59 UTC

    Thanks! You've really got me thinking.

    I'm going to add another couple of subroutines:

    my $pid = defined_or_die( fork(), "Can't fork: $!");
    my $pid = defined_or_call( fork(), sub{ die "Can't fork: $!"});

    The "defined_or_call" subroutine will only evaluate the second argument if the first argument is undef.

      Nice. But ... ;-)

      The next thing your users will ask for is defined_or_warn(), and if only for symmetry. It would contain nearly the same code as the final defined_or_die(), essentially a no-brainer.

      Then, many people (me included) prefer to use Carp instead of die() and warn(), and for them, having defined_or_carp(), defined_or_croak(), defined_or_confess(), and defined_or_cluck() would be natural. Other people would get mad if you always used Carp inside your module, so loading Carp should perhaps happen at runtime, or only when one or more of the four Carp-Class functions are exported.

      defined_or_call() is the generic solution for all of these cases, so you will probably end wrapping that function for all of the six warn()ing and die()ing functions.

      The only problem I see with these shortcuts is that they perhaps won't work reliably because the string argument may be interpolated too early:

      #!/usr/bin/perl use strict; use warnings; use 5.010; # only for // and say sub failing_func { $!=shift; return undef; } sub defined_or_warn { my ($value,$error)=@_; return $value if defined $value; warn $error; } sub defined_or_warn_reverse { my ($error,$value)=@_; return $value if defined $value; warn $error; } $!=1; say "error 1 => $!"; $!=2; say "error 2 => $!"; $!=3; say "error 3 => $!"; my $v=failing_func(1) // warn "oops: $!"; # should give error 1 $v=defined_or_warn(failing_func(2),"oops: $!"); # should give error 2 $v=defined_or_warn_reverse("oops: $!",failing_func(3)); # should give +error 3

      ... gives ...

      error 1 => Operation not permitted error 2 => No such file or directory error 3 => No such process oops: Operation not permitted at foo.pl line 31. oops: No such file or directory at foo.pl line 17. oops: No such file or directory at foo.pl line 24.

      ... on 5.10.0 (Linux and Windows Strawberry). So, no problem here as long as the error string is on the right hand side of the maybe-undefined value.

      But who guarantees that function arguments are always evaluated from left to right? Maybe that fact is documented somewhere, I'm too lazy to search right now, and after all, it's your idea. ;-)

      Another quite obvious problem is the wrong line number in the second and third error message. Your module would have to compensate that. You will probably end re-implementing or using parts of Carp.

      CGI::Carp could also be a little problem. It uses Carp internally, but it also redefines die() and warn(), and it exports its own, modified versions of confess(), croak(), carp(), and cluck(). I think a possible workaround would be to detect a loaded CGI::Carp at runtime (if exists $INC{'CGI/Carp.pm'}?) and call its functions instead of the original ones.

      Another nice shortcut could replace //=, something like defined_or_assign(\$var,$value), or even defined_or_assign($var,$value) when the wrapper has a prototype of (\$$). For boolean shortcuts, you would again need a callback function (defined_or_assign($var,expensive_or_killing_function(...)) vs. defined_or_assign_return_value_of($var,sub { expensive_or_killing_function(...) })).

      And when you really implement defined_or_assign_return_value_of(), think about a shorter name for it. The other function names are also awfully long, perhaps you could export shorter aliases for them (def_or(), def_or_die(), def_or_warn(), ..., def_or_set(), def_or_set_rv()).

      Alexander

      --
      Today I will gladly share my knowledge and experience, for there are no sweeter words than "I told you so". ;-)

        Thank you very much. Good call on not providing defined_or_die etc. That would be a lot of maintenance headache.

        I think I will start out with just def_or and dor_call for now. Once I get copyright issues worked out at work, I'll upload to CPAN.

        Thanks!

      sub _defined { if (defined($_[0])) { return $_[0]; } else { return $_[1]->(); } } sub _or(&) { return $_[0]; } my $pid = _defined fork(), _or { die "Can't fork: $!" };

      I'll let you pick better names.

      Update: Nevermind. If you try to pick useful names, you end up back with

      my $pid = defined_or fork(), sub { die "Can't fork: $!" };

Log In?
Username:
Password:

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: note [id://868433]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this?Last hourOther CB clients
Other Users?
Others having a coffee break in the Monastery: (5)
As of 2024-03-19 11:34 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    No recent polls found