On Sun, Sep 11, 2011 at 2:24 PM, Scott Godin <
sgodin@xxxxxxxxxxxxxxx> wrote:
> Hi All,
> Repro currently get's its configuration from a combination of two locations:
> 1. Command line options.
> 2. Berkely DB or MySQL datastore.
> As repro has matured the command line options list has grown significantly.
> I've heard feedback on a number of occasions that people in general prefer
> a configuration file as opposed to command line options. A configuration
> file can be fully documented, and provides some persistence of settings
> between runs (without needing to create a startup script, or batch file).
> I am in the process of converting the current command line options to be
> able to read from a configuration file instead. I was originally
> considering an approach that would allow you to specify a setting using
> either method, however the further get along with this, the more I'm
> realizing that this will just increase the complexity when adding new
> settings in the future. Besides being backwards compatible with older repro
> releases, I am not seeing any major reasons to keep the command line options
> interface around.
> Users: Would anyone be opposed to a complete deprecation of the command
> line option interface, given a configuration file interface as a
> replacement? Note: I've attached a sample configuration file to help you
> visualize how this would look.
> Developers: I'm also going to cleanup how the settings are passed around to
> the various Processors (Monkey's, Lemur's and Baboons), so that modification
> of the Processor constructor isn't required every time a new setting is
> needed. All these changes will make it much simpler to add new settings and
> use them, without needing to make modifications to 5 or 6 files.
> Scott