On 12.01.2009 18:59, William A. Rowe, Jr. wrote:
Rainer Jung wrote:
Does it make sense to add the reliable piped logs way of handling
loggers to the error log? Or is there any known dependency between the
code used in ap_open_piped_log() and the error log itself?
When I was looking at this, the difference didn't make a whole lot of sense
to me, but I was not trying to make any mass changes, only more subtle
changes that corrected the number of open and lingering fd's. Provided
that your patch to be consistent in registering reliable restart info
does work as a per-host error logger and the global stderr logger, I'd
say go ahead for 2.3 alpha.
I'll see how far it gets. Your comments and Joe's indicate, that it's
worth investigating.
But it would be interesting to look at this historically, why error loggers
were not in the original reliable piped logs implementation code path?
I just checked, and the same construction and difference between access
and error log was already in 1.3.0. From the changes, reliable piped
logs have been introduced in 1.3b2, but I didn't check that code version
to see, if it looks the same. The changes file doesn't have an entry
about error log changes w.r.t. the piped logging between 1.3.b2 and 1.3.0.
What can be seen from the CHANGES file, is that the logging of some
modules that also allow the use of piped logging, has been switched to
the new implementation only in later versions. So it's possible it was
done only for access from the start and whenever someone stumbled over
another case of piped logging, it was switched to the reliable pipes.
Regards,
Rainer
