As you might surmise, porters are generally pretty good perl programmers, and not strongly motivated by selfish needs to fix up diagnostics that they very rarely look at.1. Make a *complete* check on perldiag.pod and propose a patch. My first scan shows me that perldiag.pod is hopelessly out of sync example: lt09:/pro/3gl/CPAN 126 > grep -r "t resolve method" perl/ perl/pod/perltoc.pod:use, Cannot resolve method `%s' overloading `% +s' in package `%s', Constant perl/pod/perldiag.pod:=item Can't resolve method `%s' overloading ` +%s' in package `%s' perl/pod/perl5004delta.pod:=item Cannot resolve method `%s' overloa +ding `%s' in package `%s' lt09:/pro/3gl/CPAN 127 > Where is that message issued? The check is to be made both ways. - What messages are issued and not in diag - What messages are in diag but never issued - What messages are in diag and issued, but not in sync
This illustrates the need, and suggests a nice opportunity, for many monks to join in chant, and to commune together on the virtues of a complete perldiag, maybe even with code-snippets (aka tests) that cause those errors. This is a hard/tedious job, there are about 1000 missing entries in perldiag, so it *really* needs to be a community effort. I think it needs a bit of discussion wrt how to do it too.
1. Ive written a script to find code (Perl_warner, Perl_croak) where errors are issued, and bashed it against the contents of perldiag.pod. Id love to post it here, and see if anyone wants to use and/or improve it.
2. Ive taken its output (perldiag.pod.new), grepped out the =items, prepended each with a '#', and added a 'use Test::More 'no_plan'; to it, and called it 'diag.t'. Now it needs *lots* of examples.
3. I need a good wiki provider, where I could put a few files ( the script, perldiag.pod.new, diag.t) up, and open them up for team editing. being able to attach files is good too, esp for source code, which is easier to work on your own box, and supply patches.
4. the good is the enemy of the perfect.
HM Brand wants perfect (per the quote), but perldiag hasnt even gotten any goodness in 3 years, and Id wager the pumking would tolerate a bit of churn on that file, esp with fossilization as the alternative.
5. If you want to start putting stuff here; snippets or explanations for diags you know are missing, feel free to do so.
6. theres good reason to consider splitting perldiag into several files (peeling platform specific errs to other files is a candidate), but I think thats premature. OTOH, if you want to hack at diagnostics.pm, thats ok too. just be mindful that the hard/tedious work must be more prized.
In reply to perldiag upgrade by kidongrok
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |