Hi, On Tuesday, 17. January 2006 20:59, [EMAIL PROTECTED] wrote: > I'm a bit too lazy for cmdline tools, however using /repos/ instead of > /wsvn/ also works for me, might be a hint in the project FAQ, creating a > diff is suddenly a breeze and so is keeping up with your versions ;-) > > There's a bug in your version 0.14 which causes substring match with > schema to fail , it is fixed by the attached patch.
Thanks for spotting the bug. Whole patch committed to SVN. You may now try "svn update" or its graphical equivalent ;-) > The regex match works, however its not very common for clients to > specify regexes (we might specify a new match for that ;-), I was > actually looking for something which matches things like > schmidt schmit schmitt etc... therefore I added a list of modules which > will be tried (String::Approx, Text::Metaphone, Text::Soundex in that > order) , if you have none it will fall back on the original regexp. I guess specifying a new match is not possible. What we might try is to convert a match a la (sn~=Schm*dt) to a regex (i.e. add a . before the *) but that might get very tricky with the subtleties of regexes and LDAP filters. So be better leave it the plain regex we have now. I am not absolutely happy with the way how the approx modules are selected. I can imagine cases where I have all three installed but want to do approx matching using metaphone. (or even using regexes ;-) Maybe you can use the idea we used in Authen-SASL (get it from the same source as perl-ldap SVN ;-) There we used the import() routine of the module to choose the backend. Using this idea for Net::LDAP::FilterMatch might allow to write use Net::DLAP::FilterMatch qw/Text::Soundex/; to use Text::Soundex as approx matching backend. What do you think of it ? CU PEter -- Peter Marschall eMail: [EMAIL PROTECTED]
