[RFC] A differnt options system

Josh Sled jsled at asynchronous.org
Wed Feb 16 20:44:40 EST 2005


On Wed, 2005-02-16 at 20:24, Derek Atkins wrote:

> scheme sort of has this dichotomy that merges the concept of code and
> data.  It's sorta the nature of the beast.  *shrugs*

Yah yah, I'm well aware of that... the problem, though, is that that
flexibility comes with a price.  Since the configuration is _anything_
you can never make simplifying assumptions about it.  E.g., setting a
boolean configuration option could end up doing a distributed
transaction across a set of resources spread across the wider
internet... or maybe it'll just invoke `(lambda () (quit))`.   Which
makes no sense.

Or not.  It's only the nature of the beast because our
option-data-format is eval'ed lambdas instead of simply data.

In code, it's useful to have code be data.
In data, it's useful to have data _not_ be code.

...jsled

-- 
http://asynchronous.org/ - `a=jsled; b=asynchronous.org; echo ${a}@${b}`


More information about the gnucash-devel mailing list