Bug #25900 for Getopt-Long: Getopt-Long 2.36 breaks Verilog::Language

Thu Mar 29 00:07:26 2007 ANDK [...] cpan.org - Ticket created

CC:	WSNYDER [...] cpan.org
Subject:	Getopt-Long 2.36 breaks Verilog::Language

Let me first say, I cannot decide if Wilson uses Getopt::Long correctly. I hope the CC to him goes through and he comments on this ticket. The code that breaks is in the file vppp in WSNYDER/Verilog-Perl-2.372.tar.gz. It is being tested in t/80vppp.t. Test number 6 fails with Getopt-Long-2.36 and succeeds with Getopt-Long-2.35. This was already true for Verilog 2.371 but I did not report because I did not detect the connection to Getopt::Long until yesterday. Now I can replay: after installing G:L 2.36 this test fails and after installing 2.35 the test succeeds. So far I have verified this behaviour with bleadperl recent maintperl versions and with 5.8.8. Thanks,

Thu Mar 29 04:51:51 2007 jv [...] cpan.org - Correspondence added

From:

JV [...] cpan.org

Show quoted text

> Let me first say, I cannot decide if Wilson uses Getopt::Long > correctly.

He doesn't :-(. In Getopt::Long 2.36, the first argument that is passed to a callback is no longer a scalar containing the name of the option, but an object that stringifies to the name of the option. vppp uses Getopt::Long with callbacks, and in the code for the "simple" option it has: if ( $_[0] ) { ... push options ... } Apparently, this is intended to catch the "simple" versus "nosimple" cases. However, the option value is $_[1], not $_[0] (which, as said, is the option object). So the code should read if ( $_[1] ) { ... push options ... } Why the boolification of the callback object is not true is still a mistery for me. Besides this, vppp uses Getopt::Long in a dangerous way. It uses the pass_through config option which will effectively throw away every command line argument that is not understood. This is asking for hard to find errors.

Thu Mar 29 04:51:52 2007 The RT System itself - Status changed from 'new' to 'open'

Thu Mar 29 04:53:01 2007 jv [...] cpan.org - Status changed from 'open' to 'rejected'

Thu Mar 29 13:21:21 2007 wsnyder [...] wsnyder.org - Correspondence added

Show quoted text

>vppp uses Getopt::Long with callbacks, and in the code for the >"simple" option it has:

... Show quoted text

> if ( $_[1] ) { ... push options ... }

Indeed, I should have used $_[1] sorry about that. Show quoted text

>Besides this, vppp uses Getopt::Long in a dangerous way. It uses the >pass_through config option which will effectively throw away every >command line argument that is not understood. This is asking for hard >to find errors.

Cut and paste error. Not surprisingly I never tested --bogus flags. Thanks for noticing. I'll try to push out a new version tomorrow, I have a few other edits to collect. Feel free to move this ticket to me if you wish. BTW, thanks so much for the useful Getopt::Long and of course debugging another's bugs!

Thu Mar 29 13:21:22 2007 The RT System itself - Status changed from 'rejected' to 'open'

Thu Mar 29 14:54:10 2007 andreas.koenig.7os6VVqR [...] franz.ak.mind.de - Correspondence added

CC:	ANDK [...] cpan.org
Subject:	Re: [rt.cpan.org #25900] Getopt-Long 2.36 breaks Verilog::Language
Date:	Thu, 29 Mar 2007 20:53:38 +0200
To:	bug-Getopt-Long [...] rt.cpan.org
From:	andreas.koenig.7os6VVqR [...] franz.ak.mind.de (Andreas J. Koenig)

Show quoted text

>>>>> On Thu, 29 Mar 2007 13:21:22 -0400, " via RT" <bug-Getopt-Long@rt.cpan.org> said:

Show quoted text

> BTW, thanks so much for the useful Getopt::Long and of course debugging > another's bugs!

Johan++ and Wilson++. Thank you very much. -- andreas

Tue Jan 01 12:20:49 2008 jv [...] cpan.org - Status changed from 'open' to 'rejected'

Tue Jan 01 12:20:50 2008 jv [...] cpan.org - Fixed in 2.36 added