> I'd appreciate it if anyone who may be interested in a pure-perl > DBI emulation could explain why. Just so I can get some idea of the > real-world issues that a pure-perl DBI emulation might address.
I'd like to be able to make use of the 'use lib' function and package DBI::Lite (or whatever it's going to be called) with programs distributable on many platforms (Linux, Win32, Solaris, etc.) that are run by end users on webservers with virtual host accounts. I've done a preliminary study of various servers at random and found that DBI is installed on only half of the boxes out there. That seriously limits the distribution potential for any pre-packaged program that dreams of using DBI functions with SQL syntax to access data. For my case in particular, I would like the ability to issue DBI calls to flatfiles (CSV, Pipe, whatever) and would allow users to hook into the other DBDs at their option if the modules were pre-installed on the server. The concern is not getting any one DBD running (other than DBD::CSV or DBD::AnyData), rather getting DBI functioning to begin with. The benefit to this approach is (1) allowing the end user's to select their data format, (2) allowing the application to install anywhere, and (3) utilizing SQL to allow for easy code modification by end users. Everyone is tired of reading in flatfiles with @lines = <FILE> then splitting the lines out. Giving programs relational db capabilities that didn't have them before is a giant leap for all of us on the net. > I'd also like to know if those real-world issues could be > better addressed by bundling the DBI with perl, so if you have perl > installed anywhere then you'll also have the DBI installed there. Going forward, packaging DBI with Perl would be great, but for backward compatibility (from the existing server install standpoint) a module like DBI::Lite would make all programs utilizing the DBI functionality much more marketable to users. Thanks for letting me chime in :) Nick Hendler [EMAIL PROTECTED] _________________________________________ From: Tim Bunce <[EMAIL PROTECTED]> Date: Friday, March 22, 2002 Time: 11:10:24 AM Subject: Pure-perl DBI emulation - what use? I'd appreciate it if anyone who may be interested in a pure-perl DBI emulation could explain why. Just so I can get some idea of the real-world issues that a pure-perl DBI emulation might address. I'd also like to know if those real-world issues could be better addressed by bundling the DBI with perl, so if you have perl installed anywhere then you'll also have the DBI installed there. Tim. On Fri, Mar 22, 2002 at 08:05:59AM -0700, Troy Sniff wrote: > I'd like to get a copy of the DBI::Lite to look at. > > I might have a project that would be perfect for it. > > Troy > > > -----Original Message----- > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] > > > > I'd like to see a copy too. Sounds quite a bit like a workaround I > > was tooling around with in my head. Thanks. > > > > _________________________________________ > > > > From: Jeff Zucker <[EMAIL PROTECTED]> > > Date: Thursday, March 21, 2002 > > Time: 2:45:41 PM > > Subject: Use Lib > > > > Bart Lateur wrote: > > > > > > On Wed, 20 Mar 2002 18:16:39 -0500, [EMAIL PROTECTED] wrote: > > > > > > >DBI is just being untarred and put in the modules dir. No make or > > > >anything like that. It's for a program that will run on > > > >Linux/Unix/Windows and I don't want to have to require the compiling > > > >or installation of DBI with the program. > > > > > > > >It may be hopeless... > > > > > > It is. > > > > > > You might get by by precompiling the module for each > > platform you want > > > to support, tar up that tree, and distribute that. Pretty > > much like how > > > modules for Windows get distributed... > > > > I hesitate to suggest this. TIM, please tell me if this is too evil to > > see the light of day: > > > > I have a module which might be called DBI::Lite (or ... DBI::Emulation > > or AnyData::DBI::Lite or ?) which provides a DBI emulation in pure > > perl. Basically, you'd need DBD::AnyData, SQL::Statement, > > and DBI::Lite > > (all pure perl, no compilation required) and you could use > > do(),prepare(),execute(),fetch() and a few other features including > > placeholders. The module would work exactly like DBD:AnyData and very > > similarly to DBD::CSV but would not require installation of DBI. Once > > DBI is installed, the scripts would operate identically with > > a change of > > "use DBI::Lite;" to "use DBI;". Most users would be better off going > > straight to real DBI but in a case like Nick's it would save making > > compiled versions for multiple platforms. > > > > So, Tim, which namespace, or shall I keep it in the Dev::Null > > namespace? > > :-) > > > > -- > > Jeff > > > > >
