[GRASS5] g51 Vect_set/get_header

[Eric G. Miller]

On Tue, Apr 30, 2002 at 11:22:00AM +0200, Radim Blazek wrote:
> On Tuesday 30 April 2002 04:06 am, Eric G. Miller wrote:

> > Though, I'm not sure what wrong with just having a bunch of functions
> > to get/set things and then hide the details of the Map_info struct
> > via a typedef.  Then:
> >
> > /* Vect.h */
> > /* Error codes */
> > #define VERR_NONE    0x0000
> > #define VERR_EMEM    0x0001
> > #define VERR_EINVAL  0x0002
> > #define VERR_ERANGE  0x0004
> > /* ... */
> >
> > typedef struct Map_info * Map_info_ptr;
> >
> > int Vect_log_errors (FILE *, int, const char *);
> > int Vect_open_old (Map_info_ptr *, const char *, const char *);
> > int Vect_close (Map_info_ptr *);
> > int Vect_get_num_lines(Map_info_ptr);
> > const char * Vect_get_organization(Map_info_ptr);
> > int Vect_set_organization(Map_info_ptr, const char *);
> > ...
> 
> Sorry, I don't understand much purpose of Vect_log_errors(), should be new
> function? Which errors should log?

I didn't think it out much.  Basically, if we used an error code
reporting mechanism where the errors could be OR'ed together, then
one might want a convenient function to report each.  Don't pay much
attention to it...

> Another question: in grass50 are some functions (public - called by modules)
> named Vect_* some V1_* and some for level 2 V2_*. But for example 
> Vect_get_area/isle_points() doesn't start V2_ and works only on level 2.
> Should be this keept in 5.1, or should all non internal functions start 
> Vect_*? I prefer the later. In that case we have to rename:

I agree here.  Should be enough information in the Map_info structure
for these functions to either dispatch the appropriate function or
signal an error.

-- 
Eric G. Miller <egm2 at jps.net>