[pubcookie-dev] Re: threadsafe debug levels?
ryanc at cac.washington.edu
Thu Oct 30 12:43:12 PST 2003
No, the meaning of 0 hasn't changed. It still behaves mostly as it did, with the exceptions:
- Errors will be logged in the application log under pubcookie instead of the system log under W3SVC.
- Events from all modules (libpubcookie, security_legacy, etc.) can now make it to the log.
Below are the levels, their #defines in the filter, and their #defines in libpubcookie. Note that PBC_LOG_ERROR maps to LOG_WARN. This will make errors from libpubcookie et. al. show up as warnings in the application log. This is intentional. If one of those errors is a showstopper, which they often are not, there will be a subsequent error event from the filter.
-1 : LOG_ERR : : Errors only, This is only defined in for the Windows modules.
0 : LOG_WARN : PBC_LOG_ERROR : Warnings and Errors. This is the default and mimics the old "off" behavior
1 : LOG_INFO, : PBC_LOG_AUDIT : Informational messages. This is the old "on" setting.
2 : LOG_DEBUG : PBC_LOG_DEBUG_LOW : Stuff only a programmer would love.
3 : : PBC_LOG_DEBUG_VERBOSE: Really obscure programmer meditation. Not used in filter.
4 : : PBC_LOG_OUTPUT : Not sure. Not used in filter.
Setting Debug_Trace to 4 or higher will log all possible messages.
As for the event log message types, LOG_ERR shows up as errors, LOG_WARN as warnings, everything above that shows up as informational.
----- Original Message -----
From: "Nathan Dors" <dors at cac.washington.edu>
To: "Ryan Campbell" <ryanc at cac.washington.edu>
Cc: "Pubcookie Developers" <pubcookie-dev at u.washington.edu>
Sent: Tuesday, October 28, 2003 11:42 PM
Subject: threadsafe debug levels?
> On Sat, 18 Oct 2003, Ryan Campbell wrote:
> > I'll clean up the debugging a bit more and perhaps
> > make it less chatty at debug=1 now that we have
> > multiple debug levels available.
> How many levels are there now and what's the meaning
> of each?
> You implied that level 0 would still log errors, is
> that a change? Previously, it was a boolean and 0
> meant off. Have we changed the semantics of zero?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the pubcookie-dev