On Wed, 19 Sep 2012 17:21:32 +0200, Jens Rehsack <[email protected]>
wrote:
> Hi Merijn,
>
> while hacking around in DBD::File and DBI::DBD::SqlEngine I stumbled
> over a major design fault made in the past:
>
> sub DBD::File::Table::get_table_meta () ... evaluates
> $dbh->{f_meta}{$table}{initialized} and does something magic else. This
> magic is fully DBD::File addicted (heavily relies on file2table) and it
> should be broken into separate pieces to differ between initialisation
> done for DBI::DBD::SqlEngine and DBD::File and DBD::DBM ...
IIRC this was what you already advertised on your first round of
DBD::File internals redesign.
As long as "users" of DBD::File are not harmed, go ahead.
I do not want a truckload of code like
http://repo.or.cz/w/DBD-CSV.git/blob/8d7f4284:/lib/DBD/CSV.pm#l90
which has now been removed as the prereq's are higher:
PREREQ_PM => {
"DBI" => 1.614,
"DBD::File" => 0.40,
"Text::CSV_XS" => 0.91,
"SQL::Statement" => 1.33,
"Test::More" => 0.90,
"Encode" => 0,
"charnames" => 0,
},
But in future upgrades/updates, code like that is not unlikely to
re-appear
> I'd like to discuss it tomorrow in IRC (but I read my e-Mail if you have
> comments at the evening).
>
> If anyone else has ideas - please feel free to speak (but primary
> restriction is backward compatibility to avoid breakage of dependent DBD's).
>
> Best regards,
> Jens
--
H.Merijn Brand http://tux.nl Perl Monger http://amsterdam.pm.org/
using perl5.00307 .. 5.14 porting perl5 on HP-UX, AIX, and openSUSE
http://mirrors.develooper.com/hpux/ http://www.test-smoke.org/
http://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/
