fping will now discard replies, if they arrive after the defined timeout
for reply packets, specified with -t. This change is relevant only for the
count and loop modes, where the measured times should be now more
-consistent (see github issue #32 for details).
+consistent (see github issue [#32][i32] for details).
To prevent loosing reply packets because of this change, the default
timeout in count and loop modes is now automatically adjusted to the
## New features
-- Unified 'fping' and 'fping6' into one binary (#80)
+- Unified 'fping' and 'fping6' into one binary ([#80][i80])
- Long option names for all options
- IPv6 enabled by default
- New option -4 to force IPv4
- New option -6 to force IPv6
- Keep original name if a hostname is given with -n/--name
- Option -d/--rdns now always does a rdns-lookup, even for names, as '-n' was doing until now
-- Enforce -t timeout on reply packets, by discarding late packets (#32)
+- Enforce -t timeout on reply packets, by discarding late packets ([#32][i32])
- Auto-adjust timeout for -c/-C/-l mode to value of -p
## Bugfixes and other changes
- Option parsing with optparse (https://github.com/skeeto/optparse). Thanks Christopher Wellons!
- New changelog file format
+[i32]: https://github.com/schweikert/fping/issues/32
+[i80]: https://github.com/schweikert/fping/issues/80
(see doc/CHANGELOG.pre-v4 for older changes)