develooper Front page | perl.perl5.porters | Postings from August 2013

RFC: Should we publicize the use of locale.t for finding defectivelocales?

Thread Next
Karl Williamson
August 13, 2013 03:11
RFC: Should we publicize the use of locale.t for finding defectivelocales?
Message ID:
This is a minor matter, which I considered deciding myself or bringing 
up on irc, but in the end thought it best to get wider feedback.

The test file locale.t is now improved with more tests that look for 
improper locale implementations.  For example, it verifies that \s and 
\w are disjoint, that [:lower:] is a subset of [:alpha:], and other 
checks that a POSIX-conforming locale definition should pass.

The .t has long had reasonable debugging output (when enabled) for 
helping find out what's wrong with locales that fail.

Some years ago, it was changed so that if a platform has only up to 5% 
of its locales fail, it doesn't fail the .t, but marks those ones 
failing as TODO.  This has the advantage of not failing the build over a 
few defective locales on the platform, but the disadvantage that someone 
doesn't know that they have defective locales unless more than 5% are 
bad, unless they run the test manually.

I have just changed locale.t so it uses an environment variable to 
switch into debug mode, and to actually fail if it encounters any bad 
locales, with improved diagnostics as to what is wrong.

The question is should we publicize this, and if so, how?  It would only 
work for people who have the perl source tree or somehow could get 
t/ working.

My thought would be to add a section to perllocale.pod detailing this 
possibility and how to do it.

Any feedback?

Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About