Sam Varshavchik writes:
> Alessandro Vesely writes: 
> [...]
> > [...] Since the patch also modifies libfilter.h, it allows
> > a sysv filter to call lf_sysv_accept directly and get 0 to exit,
> > or -1 for error, as documented therein.
> 
> Your patch makes no difference as far as the rest of the Courier code is 
> concerned. 

I agree. That's also because there isn't an example of forked filter
in the standard distribution. They are mentioned as a possibility, though.

To be as clear as I can: after a few experiments, I found the most
viable solution being to

1. copy the current lf_accept code,
2. paste it into the filter
3. change its name and behaviour in order to make
 it usable with sysv semantics.

I guess any other person writing a similar filter would have to go
through the same steps. Therefore I propose the patch with the intent
to save to others the burden to do that again. It is also a more
cleaner solution, because the close-0-to-exit protocol remains
confined into the library. Hence, if you'd happen to accept the patch
in a future release of Courier, I'll happily remove my copy of the
function from my code and call the function on the library.

I'm sorry that this patch is actually causing more waste of time
than it will probably ever save to any possible forked-sysv-filter-
programmer that may happen to walk in my steps at some undefined
time in the future. So, if you're still not convinced, it's better
if I keep my copy of the function as is and we stop this confusing
thread before it mixes up our minds as it did with the subject lines.

Ale

_______________________________________________
courier-users mailing list
[EMAIL PROTECTED]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users

Reply via email to