On Sun, Jun 13, 2010 at 8:38 PM, Stanley Sufficool <ssuffic...@gmail.com> wrote: > On Sun, Jun 13, 2010 at 3:52 PM, Jared Williams > <jared.willia...@ntlworld.com> wrote: >> >> >>> -----Original Message----- >>> From: Stanley Sufficool [mailto:ssuffic...@gmail.com] >>> Sent: 13 June 2010 20:10 >>> To: Ilia Alshanetsky >>> Cc: Pierre Joye; Denis Gasparin; Matteo Beccati; >>> internals@lists.php.net; pdo >>> Subject: Re: [PHP-DEV] Re: [PDO] Re: [PHP-DEV] [PATCH] New >>> PDO methods for PostgreSQL driver >>> >>> On Sat, Jun 12, 2010 at 4:54 AM, Ilia Alshanetsky >>> <i...@prohost.org> wrote: >>> > The concerns you raised about custom methods specific to database >>> > drivers were not reflective of the PDO's intent as was >>> clarified by Wez and myself. >>> > >>> > The code that was introduced was specific to PostgreSQL, the >> common >>> > functionality was introduced in a way that allows each >>> driver to implement. >>> >>> >>> Yet the rest of the copyFrom, copyTo, etc.. could have ben >>> generic as well. I specifically stated that this could be >>> done at the driver level for FreeTDS using the BCP extensions >>> and/or using "BULK INSERT ". Yet now this is a specific >>> pgsql*** interface that cannot be abstracted for other >>> drivers OR implemented at the driver level. Now if I want a >>> MSSQL / Sybase dump/load method, I need to preface it with >>> dblibCopyFrom, dblibCopyTo. >>> >>> IMHO, every function should be a generic named interface to >>> the driver or statement level and then throw a "not >>> implemented" if the driver does not support it. This will >>> allow for abstraction at a higher level and give the driver >>> developers an opportunity to implement it. It will also give >>> PDO users a more structured way to deal with unsupported >>> methods with try/catch. >>> >>> I know I am a noob to the PDO development process, however I >>> have used OO for quite awhile and I have never seen it DB >>> abstractions implemented like this. >> >> >> But a complete nonsense for a PDO driver maintainer. >> If adds one function to a driver, then you have to had add an function >> to all the rest. > > The interface would start out as "unsupported" / NULL function pointer > in all other drivers and PDO should throw an unsupported as is does > for all the other non-implemented driver interfaces. > >> >> A generic copyFrom or copyTo method should only be added when there is >> enough PDO drivers that behave in exactly the same manner, and to do >> that, you needed specific implementations to test and compare with. >> >> If the dblib methods can't behave the same as the postgres methods, >> then clearly they shouldn't be named the same. > > Yes, DBLIB behaves the same as pgsql in that it performs the copy in a > transaction state and also has it's data visible from the server > perspective. So would SQLite 2.x "COPY", IBM "EXPORT", MySQL "SELECT > ... INTO OUTFILE", etc...
Errr... correction, BCP is client side data. > >> >> Jared >> >> >> > -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php