Marcus Börger wrote: > At 14:04 10.12.2002, Christophe Sollet wrote: > >Please let me disagree : > > > >20828 is about a bug of new locking scheme with nfs > >20858 is not bogus nor a duplicate : letting dba_open managing lock by > >default BREAK current scripts : > > > >If db was meant to be opened read only (by all httpd process) and > >filesystem have appropiate rights for that (read only), automatic locking > >will fail (and dba_open too by the way) > > > >I have undestand that by adding the new "-" flag, it will behave as > >previous release. > >But why breaking BC ?? > > > >By enabling it you may fix existing bugged php script but you may also > >break working ones. > >I think it would be better to fix bugged scripts by adding a "l" or "d" and > >keep BC. > > > >Christophe. > > > > I decided to have locking as default to force users to think about locking.
Ok, valuable whish. But what's about users that have already think about it and have implement their own locking system or don't need it (see below) ? > > Even when you can access a db file only in read mode by php locking > is required if any other process may access that file in write mode. Yes, but in our case, the db file is updated by another mean from time to time with a restart of the server : never opened read-write by any process (php or others). Again, the problem can be avoided using "-" and having php able to manage locking is great but i can't understand why it's better to break things in first place. > > Further more dba is a superset of db and db was considered to always > lock a db file (even though this is done wrong). > > And i will look at the NFS part later today or tomorrow. > Great. > > marcus Christophe. -- PHP Development Mailing List <http://www.php.net/> To unsubscribe, visit: http://www.php.net/unsub.php
