Looks good to me, other things to consider:
format_oid.h should go into oid_resolv.h?
I agree.
The functions in packet-snmp.c:
int oid_to_subid_buf()
gchar * format_oid()
void new_format_oid()
I will start and you can continue. :)
Perhaps belongs there to?
And there is also
http://www.ethereal.com/lists/ethereal-dev/200510/msg00178.html
I think oid_resolv.c should contain only name resolution.
Relation between OIDs and dissectors is another big topic which we should think about. :)
But everything does not have to kome at once :)
Brg
Anders
-----Original Message-----
From: ethereal-dev-bounces@xxxxxxxxxxxx
[mailto:ethereal-dev-bounces@xxxxxxxxxxxx] On Behalf Of Tomas Kukosa
Sent: den 6 december 2005 08:23
To: Ethereal development
Subject: Re: [Ethereal-dev] OID name translation from MIB:s
Hi,
my idea for first step is following:
1) create oid_resolv.c/h with following functions
const gchar *get_oid_name(const guint8 *oid, gint oid_len);
it would look into oid_table as weel as into MIB database
void add_oid_name(const guint8 *oid, gint oid_len, const gchar
*name);
void add_oid_str_name(const gchar *oid_str, const gchar *name);
gboolean oid_resolv_enabled(void)
it would be TRUE and later we can inroduce preference
2) Move oid_table from packet-per/ber.c to oid_resolv.c
use add_oid_str_name() in packet-per/ber.c
3) remove OID name resolution from packet-per/ber.c and call it from
proto.c
case FT_OID:
bytes = fvalue_get(&fi->value);
name = (oid_resolv_enabled()) ? get_oid_name(bytes,
fvalue_length(&fi->value)) : NULL;
if (name)
ret = g_snprintf(label_str, ITEM_LABEL_LENGTH,
"%s: %s (%s)", hfinfo->name,
oid_to_str(bytes,
fvalue_length(&fi->value)), name);
else
ret = g_snprintf(label_str, ITEM_LABEL_LENGTH,
"%s: %s", hfinfo->name,
oid_to_str(bytes,
fvalue_length(&fi->value)));
if ((ret == -1) || (ret >= ITEM_LABEL_LENGTH))
label_str[ITEM_LABEL_LENGTH - 1] = '\0';
break;
Any other ideas?
Regards,
Tom
Anders Broman wrote:
Hi,
I have checked in code to translate OID names from SNMP MIB:s if not
found in the OID names data base (Should that code be included in
oid_to_str() instead?).
This also raises the question on how to handle MIB:s.
Currently we include the ones from the SNMP library. How should we
add other MIB:s to the distribution if we want to? Another option is
to have a separate downloadable set of MIB:s.
Should we have our own "MIB" with translations we want? Should that
replace the "names data base"? We could read that file if the SNMP
library isn't available.
Brg
Anders
