[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: RoadMap Question
- Subject: Re: RoadMap Question
- From: Luke <..hidden..>
- Date: Tue, 29 Jan 2008 15:36:14 -0500 (EST)
On Tue, 29 Jan 2008, Chris Travers wrote:
> On Jan 29, 2008 8:39 AM, Paul Wrightson <..hidden..> wrote:
>
> > Both of the examples cited here are specific-ish customizations for
> > specific-ish businesses. What I mean be "specific-ish" is that the
> > functionality does not apply to the majority of businesses, nor just a
> > single business, but a sub-set of all businesses.
>
>
> >
> > What I believe we need is a framework that allows user-specified
> > workflow to be triggered under certain circumstances. For example, a new
> > customer added could trigger an external process that checks the billing
> > and shipping addresses and also checks an external credit database to
> > apply the appropriate credit limit to the account and sends a 'welcome'
> > email. Another example could be to automatically build a vendor PO
> > whenever inventory falls below the re-order value - triggered by either
> > a purchase or by a stocktaking function within LS.
>
>
> That is a good description of the problem. I think the solution involves a
> two-dimentional modularization of the software:
> 1) Vertical modularization: Stable SB APIs, Workflow scripts, and
> templates. This allows people to customize the workflow/UI of the software
> relatively easily without having to be Perl gurus or have the QA issues that
> currently affect the legacy codebase. All new or re-engineered code
> in 1.3follows the above approach. Also the new menu structure allows
> one to add
> new menu items to custom workflows reasonably easily.
>
> 2) Horizontal modularization: The current application could be broken down
> into: core/financial, inventory, and POS modules. CRM, MRP, etc. modules
> could also be written and possibly bundled. This would allow businesses who
> don't need some functionality to simply not install it.
One thing that always bugged me, was that the "API" was nothing of the
sort. It is simply a browser-free call of the CGI scripts--very
breakable, and not very flexible.
The solution that has always suggested itself to me, was daemonizing the
backend. Provide authenticated access via keys or the like.
A real command based API, which any front end--web, GUI, CLI, could be
used to access.
Yes, I know, I am out of my mind. But to me, that would be the height of
modularization.
In any case, I like what you are describing.
Luke