On Wed, Jan 19, 2022 at 02:58:34AM +0000, tslil clingman wrote:
[...] Does that make sense?
not really. ^^
Would you be able to give me some hints about how to
effect this in the current code base?
you basically want to en-loop the last "paragraph" of main() (including
the line before it). you may need to re-initialize some other members of
mvars as well, and possibly even avoid that some structures get freed
during the run - i don't remember the details (valgrind will tell you,
heh).
where it gets hairy is error handling, because it becomes really
important to differentiate between permanent and transient errors, and
the category of "permanent but acceptable" errors ("serious warnings"
from an operational perspective) exists as well. i've pondered this here
in the context of exit codes, in particular relating to a patch from
yuri.
Is there interest in such a simplistic feature upstream?
in principle yes; it adds sufficient value on its own.
my concern is to keep things reasonably clean in case a more
sophisticated trigger mechanism is added, but i don't see much room for
problems here.
the default output level should print something like "Sleeping %d
seconds before next round; use ctrl-C to exit.\n" after each round.
remember to clear the stat counters.
--periodic is a good option name, and the corresponding -p is free as
well.
base off the wip/master-next branch to avoid conflicts (the pre-existing
functionality should be still safe to use).
_______________________________________________
isync-devel mailing list
isync-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/isync-devel
