#include "mmdf.h"
struct rp_bufstruct /* for reading reply strings */
{
char rp_val;
char rp_line[256];
};
char *rp_valstr (rp_val);
rp_gval(rp_val) /* get the entire return value */
rp_gbval(rp_val) /* get the basic part of return value */
rp_gcval(rp_val) /* get the domain part of value */
rp_gsval(rp_val) /* get the specific part of value */
rp_gbbit(rp_val) /* get the basic part right-shifted */
rp_gcbit(rp_val) /* get the domain part right-shifted */
rp_gsbit(rp_val) /* get the specific part right-shifted */
rp_isgood(rp_val) /* is return value positive? */
rp_isbad(rp_val) /* is return value negative? */
Within MMDF, process, procedure, and protocol reply values are handled uniformly. The file mmdf.h defines all legal reply values and contains pseudo-functions for interpreting the values. For protocol replies, a text string is included and may be passed to the user or recorded in a log.
Values are coded into 8-bits, to allow them to be returned by Unix PDP-11 processes. A value comprises three fields. The first indicates overall degree of success. Success or failure can be complete or partial.
The second field indicates the semantic domain to which the reply refers, for example "the mail system" or "the file system".
The third field is used to distinguish the exact value.
In order to bypass machine-dependent handling of bit-strings, mmdf.h contains some macro pseudo-functions. On PDP-11 Unices, the main problem encountered is with sign-extension when the value passes through a register. The macros eliminate this problem.
The macro, rp_gval(), extracts the whole value. The macros rp_isgood() and rp_isbad() will be the most heavily used, since they test for basic success or failure.
The specific values in mmdf.h have explanations in comments.
The structure rp_bufstruct allocates a reply buffer, divided into value and string parts.