Re: new logs and errno

[Randy Terbush <ra...@zyzzyva.com>]

> I am somewhat concerned about the way some log messages are handled with
> the new logs.  Specifically, those that used to call log_unixerr.  For
> those it can be _critical_ that errno is included because that is 99% of
> the message.  Right now, that doesn't happen unless the logging level is
> set high enough.
> 
> Other messages will have completely misleading errnos in them because it
> doesn't have any meaning in that context.

I tend to agree...

I've been thinking about this over the past weeks and was mainly 
waiting to see someone else comment on this. I've been real happy 
with some of the new things I can do with the logging format 
changes, but this area is definitely weak. The only solution I have 
come up with would be to create a macro that provides a standard 
format of errno() strerror(), that can be added to the logging 
where it makes sense.

I'd welcome comments about other solutions.

Re: new logs and errno

[Dean Gaudet <dg...@arctic.org>]

Add %Z to our *printf routines which prints the strerror of the errno
value that came in at the beginning of the call (you have to save it
aside).

Dean

On Fri, 12 Sep 1997, Randy Terbush wrote:

> > I am somewhat concerned about the way some log messages are handled with
> > the new logs.  Specifically, those that used to call log_unixerr.  For
> > those it can be _critical_ that errno is included because that is 99% of
> > the message.  Right now, that doesn't happen unless the logging level is
> > set high enough.
> > 
> > Other messages will have completely misleading errnos in them because it
> > doesn't have any meaning in that context.
> 
> I tend to agree...
> 
> I've been thinking about this over the past weeks and was mainly 
> waiting to see someone else comment on this. I've been real happy 
> with some of the new things I can do with the logging format 
> changes, but this area is definitely weak. The only solution I have 
> come up with would be to create a macro that provides a standard 
> format of errno() strerror(), that can be added to the logging 
> where it makes sense.
> 
> I'd welcome comments about other solutions.
> 
> 
> 
> 
>