- loadClass() uses @ silencer on include_once..
yeah I wasnt sure how to handle this.
Generally I would make this dependent on the debug level. However I
don't know the debug level in that method.
I'd generally let the end user sort that out.. if they want to get rid of errors here the can call the method with @.. - It's a big problem with pear db, that it silently fails sometimes if you include path is incorrect.. or if you modify a driver and introduce a parse error.. - php just dies silently... - very difficult to debug..
- apiVersion() - seems a bit waste..
- why not just have var $apiVersion = 2
- isConnection = could use is_a()
- isManip() = the string is closed at the end of the line.. ALTER will
fail..
- postgres driver appears to have the same ismanip bug as DB for
EXPLAIN - see bugs db..
I just copied this over from PEAR::DB. So yeah this stuff should be
discussed and any changes should be done to both packages.
you didnt copy isManip very well :) - it's OK in DB...
- is there some reason not to follow PEAR_Errors API? - you appear to
raise errors with null messages..
Uhm. I tried my best :-)
I never really looked at the PEAR_Errors API but just used code I found.
What should I rather do?
It would be nice to hear gregs comment on implementing the 3rd parameter of PEAR_Error as an option array - (and keep BC if it's not an array..)
I guess in packages like DB/MDB, the use of a common error string for all errors is valuable... - where each of the drivers are likely to emit the same error...