Re: Help with PostgreSQL porting project - Mailing list pgsql-general
From | Chris Travers |
---|---|
Subject | Re: Help with PostgreSQL porting project |
Date | |
Msg-id | 014c01c3d135$90b3f310$1f00053d@winxp Whole thread Raw |
In response to | Help with PostgreSQL porting project ("Chris Travers" <chris@travelamericas.com>) |
List | pgsql-general |
Actually, I already had a few thousand lines of code done (using PHP's MySQL functions) and I needed an abstraction layer to conform to my existing code. In this case, if an application has been written with PHP's MySQL functions, porting it to this layer would be simple and straightforward, with a minimum of rewriting code (aside from, perhaps checking a few queries). Most of it could be handled with search/replace facilities. Even with a standard abstraction layer like PEAR or ADODB, the query syntax is not identical between database managers. Although this is not nearly perfect, I have attempted abstract over many of these differences (LIMIT clauses, timestamp formats, etc.) The abstraction is not perfect but it helps. A few additional functions could be written to better abstract some of these things. Basically these files contain a set of light-weight wrappers to facilitate porting frim MySQL to PostgreSQL of PHP applications. Basically, incompatibility comes from three things: 1: Incompatible data representation (aside from timestamps not handled by most abstraction layers, including this one). However, some of these could be better handled (such as BOOLs). 2: Incompatible function behavior (normally handled by any abstraction layer) For example PostgreSQL's return sets are not forward only, while MySQL's are. 3: Query differences. For example, if I write: SELECT * FROM customers ORDER BY last_name LIMIT 30 OFFSET 30 This will run on MySQL, but not PostgreSQL. I handled the difference in my project by adding a function to generate limit clauses for given offsets and limits. For RDBMS's that don't support offsets (MSSQL, Firebird), this would be a bit more complex, but it could be done. In other words, this is a simple API wrapper to make the db's behave the same. It is NOT OO. Best Wishes, Chris Travers ----- Original Message ----- From: "Arjen van der Meijden" <acmmailing@vulcanus.its.tudelft.nl> To: "Enver ALTIN" <enver.altin@frontsite.com.tr> Cc: "Chris Travers" <chris@travelamericas.com>; <pgsql-general@postgresql.org> Sent: Friday, January 02, 2004 7:20 PM Subject: Re: [GENERAL] Help with PostgreSQL porting project > Enver ALTIN wrote: > > > Hi, > > > > On Fri, 2004-01-02 at 13:38, Chris Travers wrote: > > > >>A few years ago, I set about porting a PHP application from MySQL to > >>PostgreSQL, after realizing that MySQL wasn't going to be able to handle it. > >>In order to do this, I built a light, fast database abstraction layer which > >>conforms to the behavior of the MySQL functions in PHP. This means that a > >>large amount of porting work could be made simple using this porting layer > >>if the application was originally used PHP's native MySQL functions rather > >>than a standard porting layer. The application was originally released > >>under the GPL. > > > > > > I guess it might bother you when people ask this but, what made you do > > the hardwork while it's already been done by several other projects like > > PHP ADODB and PEAR? > Are those two LIGHT weight? > > Afaik, PEAR (DB) affects the performance of your OO-php-app quite bad, > but I haven't really tested it very often myself, just seen tests by others? > And I'm not talking "quite bad, because php is not very fast in OO", I'm > talking "quite bad, compared to other (custom built, light weight, > simple) OO abstraction layers". > > Maybe that has changed since then? But I have, for that reason, never > really started using PEAR... > > Best regards, > > Arjen > > > >
pgsql-general by date: