> based on Theo's recent comment I thought I could take a shot at it, but
> it seems that my tree is currently broken as I can't compile it:

florian is taking lead on ping/ping6; becuase he previously did the
merge of traceroute/traceroute6.  He is quite experienced now with
the strategy this takes.

> /usr/src/sbin/ping6/ping6.c:745: error: implicit declaration of function 
> 'pledge'

If you can't compile it... then it sounds like you didn't test it well.

To hack on openbsd-current you need to be running it; recently a
number of openbsd programs were subtly refactored so that the
order-of-operations they to do is changed, to permit use of fairly
strict pledge requests.  We've taken that direction especially to
privsep programs, like ping*.

It is a new more strict way of programming unix programs (but of
course, once the results are in, someone can -D"pledge(x,y)=0" to use
the code elsewhere.

> But I thought it might still be worthwhile to share the diff.
> 
> Basically I did the following:
> 
> - renamed ping6 option -m to -D
> - renamed ping6 option -t to -o (as I would otherwise have a naming conflict)
> - renamed ping6 option -h to -t
> - renamed ping6 option -w to -O (-w is used by ping for a different matter)

That is too many steps at once, without reasoning.

Such a code-dump type of process always introduces errors.  You are
better off seperating each bit conceptually, and first discussing with
florian, who will run this show.  He'll run the show...because each of
the steps will have diff collisions against other steps, so it is
simpler for everyone if there is a strategy managed by 1 person, the
one who will do the actual commits.  That person is trying to avoid
introducing mistakes. 

BTW, I don't agree with taking simply renaming options to avoid
conflict.  I fully expect some of those options can be REMOVED,
because noone ever used them outside the KAME lab.  ping6 was not
designed in any way, it grew organically (I'll admit it, so did ping).
Each option change needs a discussion and reasoning to happen
somewhere; perhaps not out loud on a large mailing list, but at
minimum between two people... maybe 4...  1 head can easily get it wrong.


Reply via email to