Hi,
to store somewhere login name and password for each driver/database
during one GRASS session, the idea of daemon by Roger M. seems to be
the best.
http://grass.itc.it/pipermail/grass5/2002-January/001625.html
http://grass.itc.it/pipermail/grass5/2002-April/002615.html
Roger, are you still here, is your code available, can you tell us more
details: how/when is it started, how it comunicates with modules?
Or better could you contribute it directly to grass51?
Any objections or better ideas how to store passwords?
More general issue is if G_get/setenv() is sufficient for
all variables. I thing that groups of variables should be
distinguished, specific to:
- user (GRASS_GUI,...)
- session (GISDBASE, LOCATION_NAME, MAPSET, WIND?, ...)
- location?
- mapset (DB_DRIVER, DB_DATABASE,...)
- map?
It might be useful to have for G_get/setenv() possibility to set
user functions (G_get/setenv_set()) so that applications based on GRASS
(grasslib/GDAL,...) could use their own system to store variables
without conflict with GRASS files (or necessity to write to file
(G__set_gisrc_file()) )
Radim
On Wednesday 15 January 2003 02:17, Radim wrote:
Roger, are you still here, is your code available, can you tell us more
details: how/when is it started, how it comunicates with modules?
Or better could you contribute it directly to grass51?
Any objections or better ideas how to store passwords?
I'm still here.
The daemon is started near the end of Init.sh after the environment is set
up.
It communicates with modules through a pipe. It uses a simple protocol based
on some predefined messages and data packages.
The code works, but has not been widely tested. I know of two outstanding
problems. If the pipe doesn't exist when the daemon is started, the daemon
dies. The daemon allows for switching between different database locations,
but if one switches to a lat-long location the daemon dies. I think this is
just an i/o format problem, but I haven't been able to track it down.
More generally, if the daemon dies for any reason Grass becomes useless. The
daemon has to be very stable for general use. In my installation the daemon
works very well. I use it all the time and it does make Grass easier to use.
The daemon is in good shape for a demonstration, which was its initial
intent, but it isn't package to make it easily installable. I'll take a look
at what needs to be done.
Roger Miller