[Vtigercrm-developers] Receiving data from outside sources

[Allan Bush]

Why all the hate for adodb's xml database generation?

It's designed to solve all the problems you seem to want to address. 
It generates code specific to the database being used without the
developer having to worry about the different quirks of different
systems.  Also it makes updates super easy without you having to worry
about the users starting database version.

On 2/21/06, Jeff Kowalczyk <jtk at yahoo.com> wrote:
> Mike Fedyk wrote:
> > Can we keep it in adodb XML format so we don't have to have multiple
> > copies for the various database backends?
>
> That would be more elegant, but I'll suggest this practical consideration:
>
> we should ship everything as needed to install without the site-admin
> needing to run any adodb sql generation scripts, which should be the job
> of the person making the change to DatabaseSchema.xml
>
> The convention should be that the DatabaseSchema.xml is the master data
> model, and the .sql files' verbatim output are reviewed for correctness
> and committed to the repository under the exact changeset as the change to
> DatabaseSchema.xml.
>
> The .sql files defining the schema and migrations for each backend are
> never modified by developers, only generated by some adodb tool.
>
> If we don't like the generated SQL output, change the model or improve the
> tool. Or set the minimum adodb dependency version to one containing the
> upstream fix you require.
>
> This should be our policy for a number of reasons;
>
> - adodb sql generation probably works reasonably well, but we can't trust
> its cross-platform sql-generation for every subsequent adodb release.
> adodb could have bugs from time to time, and vtigercrm production
> databases will be very high-value assets.
>
> - adodb schema SQL generation output must be reviewed by experienced
> vtigercrm developers, and be invariant per repository revision per
> backend. (i.e. the .sql file output is part of the release)
>
> - site-admin deployers may not have PHP or SQL schema eyeball-verification
> experience. They may only have .sql file loading skill to the backend of
> their choice. That should be enough skill to deploy vtigercrm, and we
> should bear the repository and development overhead to make it so.
>
>
> It should be said that I don't know how DatabaseSchema.xml becomes .sql
> file output, either :)  I just remember how this has to work from my
> ADO.net XSD schema experience.
>
> When it comes to adodb schema, ask Mike Crowe ;)   Where is he these days?
>
> _______________________________________________
> This vtiger.com email is sponsored by: Zoho Writer. Are you still using your desktop word processor for typing documents? Try the AJAX enabled, collaboration-friendly online word processor, Zoho Writer for FREE instead!
> http://zohowriter.com/?vt
>