This and other RFCs are available on the web at http://dev.perl.org/rfc/ =head1 TITLE Compilation: Remove requirement for final true value in require-d and do-ed files =head1 VERSION Maintainer: Damian Conway <[EMAIL PROTECTED]> Date: 7 Aug 2000 Last Modified: 25 Sep 2000 Mailing List: [EMAIL PROTECTED] Number: 55 Version: 3 Status: Frozen Frozen since: v2 =head1 ABSTRACT This RFC proposes that files compiled via a C<require> no longer be required to end in a true value. =head1 DESCRIPTION It is proposed that the final value in a file that is compiled using C<require> no longer be significant. Instead it is proposed that files that wish to fail during compilation should throw an exception. Furthermore, any valueless exception (i.e. thrown with a simple C<die;>) that propagates through a C<require> should automatically take the appropriate message string: "require failed: file "%s" threw an exception" "do failed: file "%s" threw an exception" Note that exceptions with a value would be passed through unchanged, allowing a compiled file to signal exactly why it failed. Note too that, if the calling module wants to abort when a C<require>'d file returns a false value, it is still free to do that. The 'module initialization' feature is little-used. 99 the of 102 files in Perl 5.6 lib/*.{pl,pm} end with C<1;>. AnyDBM_File invokes 'die' explicitly. The only real exceptions are diagnostics.pm and timelocal.pl. =head1 IMPLEMENTATION C<require> should execute code in a file and return the result, as before, but it should not call Perl_die when the result is false. However, see below. =head1 MIGRATION In 98% of cases, no translation is necessary. The first version of the translator can ignore the issue entirely. Strategies to cover the other 2% follow: Is general, direct source translation of this feature of Perl 5 modules would probably be impossible. It's tempting to say that the translator should simply translate the last statement or block in the module from this: STATEMENT to this: unless (do {STATEMENT}) { require Carp; Carp::croak "... did not return a true value"; } However, I think that is impractical. The module might contain code that looks like this: if (something()) { return $v1; } ... $v2; In this case the 'return $v1' statement would I<also> have to be translated. In general, there might be many, many statements that would need to be translated. This would look awful. If complete coverage is desired, the best choice would probably be to introduce a new pragma, which would enable the old behavior. A translated module would begin with package Foo; use perl5 'require'; ... When this file was C<require>d, the pragma would set a flag. The C<pp_require> opcode would check the flag after compiling the file, and would call C<Perl_die> as before if the file returned a false value and if the flag was set. If Foo C<require>d any other modules, the flag would be cleared before loading them, and restored again afterwards. (That is, the flag would have file scope.) =head1 ACKNOWLEDGEMENTS Most of this proposal was written by Mark-Jason Dominus.
