[1457] in Kerberos
pepsy problems
daemon@ATHENA.MIT.EDU (Joerg-Thomas Pfenning)
Thu Jul 18 16:36:09 1991
Date: 18 Jul 91 18:50:18 GMT
From: pfenning@techfak.uni-bielefeld.de (Joerg-Thomas Pfenning)
To: kerberos@shelby.Stanford.EDU
In the process of bootstrapping Kerberos V5 I get unexpected results
with the output files of KRB5-asn.py. The (patched or unpatched!) pepsy
numbers the elements of structs or struct-names internally with
element_KRB5_X with X being an integer. The rest of the code magically
knows about the results and refers to the elements of the structs
with these numbers. At this point I get an inconsistent declaration
in KRB5-types.h and later usage in the *.c files.
Small example following:
The definition of
AuthorizationData ::= SEQUENCE OF SEQUENCE {
ad-type[0] INTEGER,
ad-data[1] OCTET STRING
}
KRB5-asn.py results in
struct type_KRB5_AuthorizationData {
struct element_KRB5_4 {
integer ad__type;
struct qbuf *ad__data;
} *element_KRB5_3;
struct type_KRB5_AuthorizationData *next;
};
but e.g. in file adat2kadat.c you find
register const struct type_KRB5_AuthorizationData *rv;
if (qb_pullup(rv->element_KRB5_2->ad__data) != OK)
which of course breaks the compilation. Have I missed something
important? Then I found a small glitch in the definition of the
rule PepsyTarget in Project.tmpl. The colon in the echo message
needs to be quoted with a backslash. Otherwise imake assumes the line
to contain another goal and produces the wrong indentation in the
lib/asn.1 Makefile.
--
Thomas Pfenning, University of Bielefeld, Technical Computer Science
E-Mail: pfenning@techfak.uni-bielefeld.de
Phone: +49 521 106 2918
Fax: +49 521 106 6328
Snail-Mail: Universitaetsstr., W 4800 Bielefeld 1, Germany
