Re: Inquiry about log timestamps

[Borden Rhodes <jrvp@bordenrhodes.com>]

Thank you for your feedback and I apologise for the delay in responding.

To answer the final question, I'm new at this so I need a little guidance on 
why a policy cannot be enforced.  When I'm on the BTS, I periodically see bugs 
filed against packages for violating Debian policy (usually regarding some 
licensing issue).  Why can't log timestamps be enforced the same way?

I'm not a very good programmer either, so maybe my approach to compliance is 
too simplistic.  Why wouldn't

#include <time.h>;

catch (exception e) {
	printf( "%s: Exception %s occured...", time(), e.toString() );
}

or however it's done in C work?  (You can tell I'm not a C programmer) I just 
don't see what value logs have if you can't link entries to events which may 
or may not have caused them.

Borden

On 5 February 2010 10:18:04 Julien Cristau wrote:
> On Fri, Feb  5, 2010 at 03:06:17 -0500, Borden Rhodes wrote:
> > Good morning,
> > 
> > I have a question regarding timestamps in Debian package logs.  One of my
> > biggest frustrations when troubleshooting problems with Xorg, for
> > example, is that the .xsession-errors and Xorg.log entries are not time
> > stamped.
> 
> It's hard (read impossible) for .xsession-errors since that's just the
> stderr output from the apps running in your session.  For the Xorg log,
> http://cgit.freedesktop.org/xorg/xserver/commit/?id=d2322b6309bf15a45002b42
> e7e6ba3d6b5bfa932 will be in xorg-server 1.8 (people have tried to get
> better timestamps in there, but since it needs to be signal safe, it's not
> easy).
> 
> In any case, I'm not sure what you think policy should say about this.
> It could make timestamped log files a recommendation, but I don't think
> it can enforce it.
> 
> Cheers,
> Julien