> How many unique strings are we translating?
Very rough-n-ready figures:
In the snmplib code, there are
92 calls to 'snmp_log*'
(not counting snmp_logging.c)
80 calls to 'snmp_perror' or 'config_perror'
~260 calls to 'printf' or 'fprintf'
(160 in mib.c or parse.c - most
of which wouldn't need translating)
In a (presumably) typical command line tool (snmpget), there are
3 calls to 'snmp_sess_perror'
10 calls to 'fprintf'
In the agent top-level code (i.e. excluding MIB modules and handlers),
there are
33 calls to '*perror'
77 calls to 'snmp_log*'
96 calls to '*printf'
That makes something of the order of 500 output messages in total
(ignoring the MIB parsing code).
That's assuming that these are all different, and all need
translating. Neither of which is likely to be true.
> Of course, if one wants to support both gettext and catgets in this way,
> the natural first step would be to add support for gettext - theese
> cludges could be added later.
That's certainly the way my mind's turning....
Dave
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
|
