Re: pearifying metabase

Hello, "Tomas V.V.Cox" wrote: > > Now, it is the time to make a proof of concept: wrap PEAR-DB API around > > Metabase and demonstrate that it works while it can be expanded to > > handle Metabase features that PEAR-DB does not yet provide. > > Well the effort should be in both ways: one for PEAR API to have the > Metabase Features and other for Metabase to be able to handle features > that PEAR DB has (apart from the API), ie: > > - Abstracted DB errors > - Some feature emulations > - PEAR Error support (setErrorHandling, pushErrorHandling, expectError, > etc) > - Some missing drivers in Metabase like Sysbase, Frontbase, LDAP > > (sorry I don't have too much knowledge of Metabase to be more verbose) > > Perhaps when Lukas presents the things that both have missing and how to > adopt them, the picture will be completed. One thing at a time. The PEAR-DB API first. > > What would help from your side is some conformance tests that you can > > run on PEAR-DB API so we can validate that once it wraps around Metabase > > it still works as intended, or else we will never be sure if merger will > > really work and may slip fundamental flaws and limitations that may be > > too late when we realize them. > > Some time ago I did a long test batery suitable for all the drivers > (based in the specific driver tests Stig did long time ago). My recent > work apart from building it is to ensure that all drivers pass it. You > can find it in the PHP repository at: > > php4/pear/DB/tests/driver/ > > Just edit the setup.inc file, put your DSN and launch the tests with: > > php4# make test TESTS=pear/DB/tests/driver > > They test almost everything, well, I will add a missing one: multiple > conections and nested queries. I don't mean tests that verify if the current PEAR-DB implementation works but things that you would like to make sure that they work using Metabase wrapped by PEAR-DB API before you feel confident that the merge will work. This needs to be done quickly before the effort that comes ahead is much greater. If there are fundamental flaws that will prevent anybody from approving the merger for once, they should be spotted now. Lukas decided to invest a great deal of time on this but I am sure he alone will not be able to spot all problems that may arise because he doesn't know PEAR-DB so well, so he'll need all the help he may get from people experienced in PEAR-DB sort all the issues before it is too late. Regards, Manuel Lemos