Re: vxlan bug wrt IN6_ANY as source Was: Re: tweak in6_selectsrc()

2016-12-02 Thread YASUOKA Masahiko
Hi, > I think we should not allow at all empty source addresses, as it can > make things confusing when troubleshooting. goda@ yasuoka@ reyk@ : > what is your take on this ? I have the same opinion of you. Since tunneling protocol like vxlan (or etherip) is used with statically assigned IP

doas.conf(5): clarify args

2016-12-02 Thread Anton Lindqvist
Make it clear that args is a keyword followed by zero or more arguments. Unfortunately, the grammar format no longer fits on a single line even with the offset removed. Index: doas.conf.5 === RCS file:

Re: vxlan bug wrt IN6_ANY as source Was: Re: tweak in6_selectsrc()

2016-12-02 Thread Reyk Floeter
> On 02.12.2016, at 11:31, Reyk Floeter wrote: > >> >> On 01.12.2016, at 08:35, Vincent Gross wrote: >> >> On Tue, 29 Nov 2016 17:03:44 +0100 >> Martin Pieuchot wrote: >> >>> Diff below removes the 'struct route_in6' argument from >>>

Re: vxlan bug wrt IN6_ANY as source Was: Re: tweak in6_selectsrc()

2016-12-02 Thread Reyk Floeter
> On 01.12.2016, at 08:35, Vincent Gross wrote: > > On Tue, 29 Nov 2016 17:03:44 +0100 > Martin Pieuchot wrote: > >> Diff below removes the 'struct route_in6' argument from >> in6_selectsrc(). >> >> It is only used by in6_pcbselsrc() so move the code

Re: Two functions for cwm

2016-12-02 Thread Erling Westenvik
On Thu, Dec 01, 2016 at 12:26:36PM -0500, Okan Demirmen wrote: > On Thu 2016.12.01 at 17:25 +0100, Erling Westenvik wrote: > > Just a proposal. I like to prevent selected programs (like xclock, > > xbiff, xload) from filling up the taskbar (tint2 in my case) and added a > > while ago a function

Re: doas.conf(5): clarify args

2016-12-02 Thread Ingo Schwarze
Hi, Anton Lindqvist wrote on Fri, Dec 02, 2016 at 10:40:16AM +0100: > Make it clear that args is a keyword followed by zero or more arguments. Your argument makes sense to me, we generally don't use "..." to imply a preceding .Ar, but we use it to indicate that the preceding .Ar can be

Re: Two functions for cwm

2016-12-02 Thread Erling Westenvik
On Fri, Dec 02, 2016 at 03:17:05PM -0500, Okan Demirmen wrote: > On Fri 2016.12.02 at 11:15 +0100, Erling Westenvik wrote: > > On Thu, Dec 01, 2016 at 12:26:36PM -0500, Okan Demirmen wrote: > > > On Thu 2016.12.01 at 17:25 +0100, Erling Westenvik wrote: > > > > Just a proposal. I like to prevent

switchd(8): learn remote switch tables

2016-12-02 Thread Rafael Zalamena
Learn remote switch's flow table properties so we can use this information to decide where to install the default table-miss flow for OpenFlow 1.3. This is not needed by OpenFlow 1.0 since it already does this by default. This diff implements the functions to ask the remote switch for tables

Re: ntpd(8): use stack instead of heap

2016-12-02 Thread Rafael Zalamena
On Sat, Oct 01, 2016 at 07:05:51PM +0200, Rafael Zalamena wrote: > The ntpd(8) constraint fork+exec diff changed the way the constraint > processes are created, but then it introduced new calloc()s to avoid > increasing diff size and to focus on the problem. Now that the fork+exec > is in, this

Re: Two functions for cwm

2016-12-02 Thread Okan Demirmen
On Fri 2016.12.02 at 11:15 +0100, Erling Westenvik wrote: > On Thu, Dec 01, 2016 at 12:26:36PM -0500, Okan Demirmen wrote: > > On Thu 2016.12.01 at 17:25 +0100, Erling Westenvik wrote: > > > Just a proposal. I like to prevent selected programs (like xclock, > > > xbiff, xload) from filling up the