bug-hurd
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Hurd logging. (was zalloc panic)


From: Jon Arney
Subject: Re: Hurd logging. (was zalloc panic)
Date: Thu, 28 Feb 2002 23:09:09 -0700

Neal H Walfield wrote:
> 
> Can you justify why this is better than syslog?

It is not inherently better than syslogd.  It does, however,
serve a slightly different class of process.  I am not opposed in
principle to adapting syslog to handle Hurd/Mach ports as
opposed to Unix  domain sockets or UDP sockets.  The current
implementations of syslogd I have been exposed to assume that
the caller is sending information out on Unix sockets, UDP
sockets which in my opionion are not the appropriate vessels
for logging Hurd events. This is particularly true if the
socket translator is to send messages. Using syslog as the
basis for an implementation is probably a good idea, but in
its present state, I don't think it is usable in general for
hurd translators.

The second point is that many of the hurd translators do not use
syslog (perhaps out of concern for the issues I raise above).
In fact, the only ones in the source base I could see were:

daemons/getty.c, 
daemons/lmail.c,
trans/pump.c
ufs-utils/newfs.c

Most of these are not real translators at all, but ordinary
processes.

On the other hand, many translators call 'error()' which according
to glibc source does an 'fprintf(stderr' and I assert that logging
may be more appropriate in some cases than a console message.  In
some cases, a simple 'errno' return value suffices which may or may
not give an indication of what REALLY happened (as in the 'auth'
example I pointed out earlier).

Just my opinion.

Jonathan S. Arney
Software Engineer
jarney1@cox.net
------------------------------------------------------------------------



reply via email to

[Prev in Thread] Current Thread [Next in Thread]