[Gpsbabel-code] Some more thoughts on architecture

[Ron P. <pa...@fw...>]

I've been thinking some more about some big-picture type things.
Now obviously, this project is Robert's baby and I'm just being
allowed to play peek-a-boo with it, so these are all just requests 
for comment, not concrete suggestions or laws or anything.

Tossing around my question about the command-line parameter has
led me into slightly different territory.  My big concern was 
that if I steal -D or -N or whatever for a specific output module,
that's one less letter that can be used for general options if
they come along.  So, it occurred to me that such a parameter
should be specified using a prefix character of some sort.  'O'
would make a decent prefix character, I suppose.  So instead of 
what I said before, I envision something like this:

 gpsbabel -i geo -f foo.loc -o cetus -On "database name" -F foo.pdb

This allows each output module to have its own set of switches,
independent from the switches of every other output module.  

Expanding on this idea, of course, you'd have -I as a prefix for
module-specific arguments to input modules.  For example, I might
choose to import only the waypoints in a particular category from
some Palm format (this is hypothetical; none of the currently-
supported Palm formats use categories, though magnav does require
a category header.)  To do that, I could use "-Ic Geocaching" to 
get just the waypoints in the "Geocaching" category.

Finally, just before derailing completely, my train of thought
wandered into the land of "coalesce" and "simplify" by way of
a thought experiment involving a hypothetical filter that could 
grab NAPP/Terraserver imagery for each waypoint for use with 
GPSPilot.  These things are all just filters; things that happen
between when the waypoints get read and when they get written.
They, too, will eventually suffer from a shortage of switch 
characters.  Why not handle them all with a single switch, or
pair of switches?  -f and -F are unfortunately taken, but we
could call them modifiers instead of filters:

 gpsbabel -i foo -f a.foo -f b.foo -m coalesce -o foo -F c.foo

merges a.foo and b.foo into c.foo, removing duplicates.

And, a really complicated example:

 gpsbabel -i bar -Ic "Geocaching" -f ${foo}.pdb \
  -m declination -m cachedesc -Mp proxy.example.com \
  -o gpspilot -On "GC Waypoints" -F ${foo}-wpt.pdb \
  -o gpspilot_img -On "GC Imagery" -Or 1km \
  -Op proxy.example.com -F ${foo}-img.pdb

reads the "Geocaching" category from the bar-format database 
${foo}.pdb, computes the magnetic declination for any point that 
doesn't already have one, fetches the geocaching.com cache 
description for each "GCxxxx" waypoint into the "long description" 
field, writes the resulting waypoints to a gpspilot waypoint 
database called "GC Waypoints," and fetches imagery for each 
waypoint to a radius of 1km, which it writes to a gpspilot 
imagery database called "GC Imagery."  (gpspilot_img would 
be invalid as an input format, obviously.)  Oh yeah, and our
web connection is through a proxy.  Simple, right?

So... any thoughts?