httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Marr <>
Subject RE: logging APR errors
Date Mon, 17 Jul 2000 19:26:27 GMT
At 12:04 PM 07/17/2000 -0700, wrote:
>On Mon, 17 Jul 2000, William A. Rowe, Jr. wrote:
> > Wrong.  If we design this correctly, the error handler is actually a 
> stack.[...]
> > again at a lower point in the stack.  Whereever we simply -don't care-
> > (and there will be many examples even within APR itself) then we can catch
> > and drop the error, or catch and transform if the error is meaningless to
> > the lowest predecessor.
>This isn't the design being discussed.  The design being discussed has us 
>adding code to APR for those cases that Apache wants more information.

No, the design being discussed has us adding code to APR for those cases 
where APR wants to report more information than can be returned with a 
status code.  As far as I can see, you're the only one tying this to Apache.

>This is not a good thing.  We would need to add code to APR for all 
>cases.  Then, Apache needs to decide whether or not it wants to log an 
>error before EVERY APR call.  There is no way around that.

I doubt very much that you could prove that.  What you probably meant to 
say is that you have not been able to think of a way around that.

>If Apache doesn't want to log an error, it has to look at the stack, and 
>push NULL, or leave the stack alone.
>If we just use a stack, and just push a function when we want to log, and 
>pop after the APR call, then we are still impacting the main-line 
>case.  The advantage to this approach, was that it doesn't impact the 
>main-line case.  This is untrue.  There is no way for a logging callback 
>function to not impact the main-line case that I can see.

This all assumes that the callback isn't intelligent enough to make the 
decision on its own whether or not the error should be logged.

View raw message