01:29:30 <occ> why can freebsd  get ipv6 address with wire network card, but can not get address with wireless device ?
01:30:00 <occ> both ways use dhcp to get address.
01:30:36 <occ> where might it be wrong ?
01:32:04 <occ> using freebsd 13.x
01:34:03 <RhodiumToad> iirc, dhcpv6 is not supported in the base system
01:34:38 <RhodiumToad> this is because ipv6 is not supposed to use it, it has its own address discovery mechanisms, and dhcpv6 is a hack
01:35:16 <RhodiumToad> what makes you think it's using dhcp when on wire?
01:36:47 <occ> sorry, I  don't know that.
01:37:57 <occ> how to setup ipv6 connect with wireless connection?
01:39:06 <RhodiumToad> I don't know, sorry
01:39:59 <occ> I searched some web pages on how to configure dhcpv6 on wireless network but I can't duplicate that.
01:43:58 <CrtxReavr> ifconfig_vtnet0_ipv6="accept_rtadv"
01:44:16 <CrtxReavr> Add that to your rc.conf, only switch vtnet0 for your wifi interface.
01:54:05 <RhodiumToad> occ: did you see CrtxReavr's response?
01:56:23 <rwp> I have yet to find a public use (coffee shop) WiFi Access Point that passes IPv6 through. Here they are all IPv4 only.
02:03:03 <occ>  RhodiumToad: I was finding more information on accept_rtadv ...
02:08:54 <occ> rwp: in my case, I happened seeing the phenomenon that freebsd could get ipv6 address with wire connection but only ipv4 address with wireless connection at the same time.   freebsd can  visit public network with above both connections.
02:09:23 <occ> s/both/two
02:11:21 <occ> two default routes work fine, one is ipv6 ,the other is ipv4.
02:11:27 <rwp> The more I learn about IPv6 the more I think DHCPv6 is the better option over SLAAC/RA
02:24:15 <occ> I didn't use ifconfig_vtnet0_ipv6 in /etc/rc.conf.
02:25:47 <occ> just ifconfig_(device)="DHCP"
02:35:43 <occ> I didn't use ifconfig_vtnet0_ipv6 like options in /etc/rc.conf.
03:21:54 <RhodiumToad> dammit, what's a good way to compare a constant like SIZE_MAX against a value without triggering -Wtautological-constant-out-of-range-compare
04:06:26 <epony> they make compilers out of warnings now
04:39:35 <rwp> RhodiumToad, Hmm...  SIZE_MAX is the largest size possible, right?  So won't by definition all values either be equal to it or less than it?  Can't ever be larger.
04:40:46 <rwp> In other words...  Does it make much useful sense to ever compare against it?
04:41:22 <RhodiumToad> yes
04:41:55 <RhodiumToad> assume you have a variable foo which is of an (assumed unsigned) type defined by some library header
04:42:14 <RhodiumToad> you want to know if you can safely cast (size_t)foo preserving the value
04:42:42 <RhodiumToad> now, size_t isn't guaranteed to be the largest unsigned type, in particular on 32-bit sustems it's not
04:43:41 <RhodiumToad> so you might try  if (foo <= SIZE_MAX)  on the assumption that whichever of the two sides is a smaller type should be promoted to the larger one
04:44:00 <RhodiumToad> on 32 bit where foo is a uint64_t, this works without error
04:44:08 <RhodiumToad> but on 64 bit, you get the warning
04:44:12 <rwp> I'll play Devil's Advocate here and say then that sizeof target needs to be >= sizeof source.
04:44:38 <RhodiumToad> sizeof isn't really the right tool to compare type ranges
04:44:59 <rwp> Well...  Neither is foo <= SIZE_MAX either.
04:45:16 <rwp> Since that must always be true.
04:45:17 <RhodiumToad> and it's explicitly OK for foo to be of a larger _size_ than size_t, as long as the actual value fits in a size_t
04:45:47 <RhodiumToad> no, foo <= SIZE_MAX is not always true since foo can be of a larger sized type than size_t
04:49:40 <RhodiumToad> the other method is  if ((size_t)foo == foo)
04:50:08 <RhodiumToad> (assuming foo's type is unsigned, this is well-defined)
04:50:51 <RhodiumToad> if foo's type is no larger than size_t then the cast preserves the value and the result is true,
04:51:12 <RhodiumToad> whereas if foo's type is larger and the value fits in size_t, the result is still true,
04:51:39 <RhodiumToad> but if foo's type is larger and the value doesn't fit, then the cast reduces the value mod 2^N for some N, and the comparison is now false
05:05:30 <RhodiumToad> but annoyingly, that version generates much worse code
06:42:09 <nmz-> how do you run du -b on fbsd?
06:52:03 <nmz-> welp, installed busybox I guess
07:51:57 <yuripv> nmz-: if it comes from gnu du, install sysutils/coreutils
09:02:03 <_xor> I have to admit, I had a "Eureka!" moment earlier today that I think is pretty cool.
09:02:20 * _xor is working on an implementation for orchestrating jails across a cluster using pkg as the image repository
09:52:28 <trev> how do i install a man page for my port?
09:53:06 <trev> do i have to do it myself in the source Makefile or can i set something in the port Makefile?
09:53:21 <yuripv> https://docs.freebsd.org/en/books/porters-handbook/makefiles/#makefile-manpages
09:53:22 <VimDiesel> Title: Chapter 5. Configuring the Makefile | FreeBSD Documentation Portal
09:55:26 <trev> thanks. i only read the quick ports part :[
09:55:57 <yuripv> it should be installed by source Makefile, yes, ports are not really a build system, rather a wrapper around original one and patching as needed
09:56:47 <trev> is there a freebsd way to do that?
09:56:53 <trev> i don't see something like `install-man`
09:57:58 <yuripv> if you use the system mk files, it's MANx= something.x, where x is the section
09:59:24 <yuripv> (ugh, or is it only MAN= something.x now, correctly parsing the .x for section)
10:00:06 <yuripv> both should work anyway :)
10:42:49 <espen_> Hi, I'm struggling with a devd rule where my action can't resolve a variable: https://bsd.to/LkHX
10:42:50 <VimDiesel> Title: dpaste/LkHX (Plain Text)
10:43:34 <espen_> sorry, better formatting: https://bsd.to/nWev
10:43:35 <VimDiesel> Title: dpaste/nWev (Bash)
10:53:40 <espen_> I guess it's more a problem with su shell and my bad shell scripting skills in general than devd :)
11:14:09 <yuripv> espen_: devd.conf(5) has a "Notes on Variable Expansion" subsection
11:16:13 <yuripv> (and i think the bottom line there is "don't use single quotes" :)
11:26:40 <espen_> yuripv, thanks! and yes you're right.. now I need to figure out how to replace those o_O
11:28:52 <espen_> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=240411
11:28:54 <VimDiesel> Title: 240411 – devd behaves odd when action uses single quotes
11:29:53 <espen_> I think I'll have to get rid of the variable and have one rule per device :/
11:30:35 <yuripv> espen_: you need single quotes for -c argument to su?
11:32:24 <yuripv> espen_: may be try using escaped double quotes in place of single ones? i.e. -c \"...\"
11:32:39 <espen_> Not necessarily I think, but I'm failing to escape it and devd accept it.
11:32:45 <espen_> Yeah that doesn't work..
12:26:59 <trev> for install(1), is it true that you can't actually do -d to create parent dirs and install files at the same time?
12:29:10 <rustyaxe> fwiw on (other) systems man install shows for -d:                treat all arguments as directory names; create all components of the specified directories
12:29:19 <rustyaxe> Which vaguely goes back to my distant memories of POSIX
12:32:00 <yuripv> istr there was a pr for it
12:32:19 <yuripv> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=230620
12:32:22 <VimDiesel> Title: 230620 – "install -d" issue
12:34:22 <yuripv> rustyaxe: the problem seems to be creating directories *and* installing files at the same time
12:34:40 <rustyaxe> yea i think posix says -d treats all args s directories
12:34:53 <rustyaxe> thus GNU's behavior and what i remember from 30 years ago :)
12:35:21 * rustyaxe pops into openbsd box and looks
12:36:14 <rustyaxe>      -d         Create directories.  Missing parent directories are created as required.  This option cannot be used with the -B, -b, -C, -c, -f, -p, or -s options.
12:36:36 <rustyaxe> im thinking htats expected behaviour
12:36:43 <rustyaxe> Deal with directories first, then deal with files
12:36:50 <trev> yuripv: yep that's it. thanks
12:37:12 <trev> i just used mkdir -p instead
12:37:16 <trev> extra step but whatever
12:37:18 <rustyaxe> if not expected, at least the most common behavior -- ie if you care of portability, dont do it
13:10:05 <trev> why would you need to have `make package` succeed as non-root if the ports dir is /usr/ports?
13:10:15 <trev> (following quick port steps)
13:18:33 <gzar> hey folks, where do i find the libc-devel package/sources? im building the linux kernel from scratch and its complaining that it cant find bytewswap.h header
13:23:31 <yuripv> gzar: wrong channel?
13:24:26 <gzar> no
13:24:31 <gzar> im on freebsd
13:28:48 <gzar> basically just looking for the libc devel files
13:31:03 <rustyaxe> i'm curious as huge as linux kernel has grown if cross compiling has even been tested in decades :P
13:33:18 <gzar> oh, well i guess i am testing it now
13:34:04 <gzar> i've been thinking of an experiment for a while with the linux kernel
13:34:34 <gzar> but will probably ask in some linux channel
16:03:49 <hubertm> if my printer (samsung M2026) has no drivers for FreeBSD, is there chance to get any support beyond ascii printing?
16:22:28 <mcrane> hubertm: cups has some generic drivers when I tried that on my printer it worked fine
16:23:05 <mcrane> so thought it didn't show my specific model it still worked when I looked around for something more generic