> Saleem Ahmed Khan wrote:
> > We have been working on squid-2.6.STABLE18 and as this been already
> > deployed, changes are required to done on this.
Hello Saleem,
If you can submit a patch for both Squid2 and Squid3, that would be
great. Otherwise, you will be at the mercy of others to update your
changes to Squid3.
> > To fulfill this requirement currently i am putting a check in
> > "parse_cachedir" function of cache_cf.c to see if the cache directories
> > specified in squid.conf are present or writable by using stat system
> > call. Based on the true condition I am allocating the swap directories
> > or returning from the function.
Ideally, we should try to separate configuration parsing and actual
configuration actions whenever possible. Doing so here may require more
work than you may be willing to do for this small change, unfortunately,
but the directory initialization failure should be detected by the cache
directory initialization code (which could be specific to the cache dir
type!). The caller (i.e., the code looping over all configured
cache_dirs and initializing them) can then catch the initialization
error and handle it gracefully if configured to do so.
Again, I do understand that I may be asking too much here. I am trying
to minimize the number of tiny features added to wrong places as it
complicates long-term code management, but I realize that blocking such
tiny features (when correct implementation is difficult) is not
pragmatic.
Thank you,
Alex.
Received on Tue Apr 22 2008 - 13:14:09 MDT
This archive was generated by hypermail 2.2.0 : Wed Apr 30 2008 - 12:00:07 MDT