just saw a generic Python app that should run anywhere, but their github page says: if you don't use the Flatpak version, don't report bugs. so they've basically taken something that should work on any Unix system and made it Linux-only for no reason at all.

how do i list service on my machine which are not started or enabled yet?

perhaps service -l | xargs -n1 -Isvc service svc status 2>/dev/null |grep -v 'is running'

ouch

does wayfare run as a service?

wayland

no, wayland itself is just a protocol, the 'server' is a compositor application like sway, hikari, gnome, kde, ...

there are some display manages that support wayland that can run as a service... i think sddm is one of those

s/manages/managers/

labwc cannot read /rc.xml

need a login manager. is lightdm good?

is lightdm a login manager?

lw: doesn't python have enough packaging options?

meena: yeah but everything has to be snap or flatpak nowadays because... uh... reasons, i guess

pipx install <python-app> # if that's not your recommended way to install your application, wait, never mind, I'm working on one of those that you can't install that way

if i'm filing a bug against clang, is there a keyword i should use?

lw: if clang or llvm exist as keywords, you can use them. but keywords are kinda underused and unwieldy on Bugzilla :(

they apparently do not, i'll just let someone triage it

"someone" (Mark)

AumShivaya: i filed a bug for the dxx-rebirth issue from the other day, but iirc this was not the same problem you were having with it... bugs.freebsd.org/bugzilla/show_bug.cgi?id=276306

Title: 276306 – building games/dxx-rebirth causes clang ICE

my issue was 1. no pkg package and 2. ports did not download distfile

oh you're on 15

yeah, 15 has a newer clang (17.x)

although i could fetch the distfile fine so i don't know what's going on there

meena: poor mark

lw: i do about 0.1% of the work Mark does, so it's not all on him.

lw: hrm… shouldn't you be submitting that upstream?

meena: i don't know what patches the freebsd clang has... the error did say to submit a freebsd PR, not an llvm one

aye

Title: LLVM / project - FreshBSD

i mean i could go and reproduce it upstream and then report it there and/or file a PR to backport the relevant upstream patch but... i am working on something else right now and don't actually care about this port, i just noticed it in passing trying to help someone else

maybe i will go do that when i have some free time if it hasn't been fixed :-)

and anyway i did one pull request and one bug today so, karma-wise, i'm even :-P

ok i got a minimal repro, seems like a freebsd bug no? termbin.com/pd9k

ok i got a minimal repro, seems like a freebsd bug no? termbin.com/pd9k

alepzi-, Did you restart the "routing" service afterward? If not then do that. "service routing restart"

rwp yep that works

shouldn't restarting netif make it restart routing if it needs that?

One might have put everything into one init script but for whatever reason I know not it is split into several parts and routing is in the routing script.

And bridge stuff in the bridge script. And so on. It's all modular.

is there no way to reload all networking related stuff with 1 service call?

ok here's the reason:

in an unattended bsdinstall it starts, expands base, expands my distros, then it runs my config script in installerconfig. well that relies on networking etc, so i wanted to 'start' the networking based on the config in my distro that was in place

it works for vm machines but this vm host has a more complicated networking setup (bridge and tap) and it's not working here

service netif restart then route add default 10.1.1.1 was enough, now it's not, so debugging that

you brought up bridge, so maybe i need to restart the bridge service too?

Probably. I don't know. I am just a newbie on this.

rwp that got it! service bridge restart in addition to netif and routing

i'm holding you in front of me like rose from titanic

fly my queen

o/~

there's a bridge service? ... oh, that's for autobridge

There are many services. hostapd, ipfw, stf, wpa_supplicant, everything is modular components.

well, yes, but bridge is not something that would usually require a service because it's configured by ifconfig. but i forgot about autobridge.

imo we need better launchd integration with rc so that restarting netif would know routing and bridge were dependencies and were also restarted

the whole rc service relationship graph could be encoded in a launchd format and be pretty cool

there is nothing like launchd/systemd/SMF on freebsd, it's just init and some scripts. perhaps it would be nice if there was a command that could restart a service and the things that depend on it, but not automatically, because that's often not required

pretty sure launchd was built for macos and freebsd at the same time like 15 years ago

might be a lil wrong on the details

i don't know what the history of launchd is but it's not in freebsd

apparently there was some (very old) discussion about merging it, and a clone called jobd: wiki.freebsd.org/launchd

Title: launchd - FreeBSD Wiki

also github.com/freebsd/openlaunchd which is dead

Title: GitHub - freebsd/openlaunchd: A launchd(8) port for non-Darwin systems

imo it would be a perfect backend lib for a rc rewrite to build on

better than the systemd hog

very light weight

Actually ifconfig is used in all of those service files. All of those each call ifconfig to do the actual work.

by 'ifconfig' i meant ifconfig_bridgeX="..." in rc.conf. i'm aware bridges are configured with ifconfig :-)

grep ifconfig /etc/rc.d/*

I was responding to the comment "bridge is not something that would usually require a service because it's configured by ifconfig" which sounded funny to my ears because all of those things are configured by ifconfig.

ok, then let me rephrase. i was surprised there's a dedicated rc script for bridge because bridge is usually configured using ifconfig_bridgeX="..." in rc.conf, which is handled by /etc/rc.d/netif.

but i had forgotten that 'autobridge' is a thing (because i don't use it) which is what /etc/rc.d/bridge does

I like that rephrasing. Has a better rhythm and flow to it. :-)

hmm, i wonder why mkvtoolnix is only managing ~400Mbps over NFS. cpu is < 50%, disks are mostly idle...

hm, i'm suspicious that HandBrake on FreeBSD is encoding a video at ~18fps when the Windows version on the same system only managed ~9fps. makes me think i configured something wrong

answering my own question from a few days ago, the best replacement for Mp3tag on freebsd seems to be audio/puddletag, because it's literally an almost identical clone of mp3tag

lw: what makes it better?

kenrap: well, the only other option i found was easytag (which someone here suggested), which looks okay, but puddletag's ui is more like what i was looking for

I see. Nice!

oh, if you mean better than Mp3tag... that only runs on Mac and Windows, so it's not a good option for freebsd :-)

although i did consider trying it under wine, might still do that if i don't end up getting along with puddletag

Yeah, why bother when there is a native alternative, and one that has a better UX too :)

what are some amazing FreeBSD games, I tried to run dxx-rebirth but it does not install nor do I have the .hog files

but there has to be some others worth playing?

does anyone know if there's a way to make it so that the geli providers created at boot (in a geliboot+zfs configuration) use the /dev/diskid names, instead of raw device names?

(and if not - should there be?)

I'm having a hard time figuring out exactly what's responsible for which gets picked

this is pre-root-mount, so it can't be rc scripts

I need to do other things with this system right now, but I'll try and figure out later whether the blame lies with the loader or the kernel

Hi all. So, according to this: churchers/vm-bhyve #525 and this: bugs.freebsd.org/bugzilla/show_bug.cgi?id=273560 - should all UEFI VMs (OpenBSD, FreeBSD, Windows) have bhyve_options="-A -P" in their conf files?

Title: Fix UEFI VM options by dag-erling · Pull Request #525 · churchers/vm-bhyve · GitHub

i am trying to port a makefile from linux. there is a `ln -srf -T <path1> <path2>` in it. the -r option from linux is confusing me.

it is something akin to: when creating a relative symlink in a different location than the current directory, the resolution of the symlink will be different than the resolution of the same string from the current directory

actually, i have just realized that the makefile is written in bash and assumes a bash shell in order to run the login manager which it installs. so all i need to do is change my .profile shell to bash after installing bash???

is this correct? how can the display manager utilize bash before the .profile is read?

you can set your user interactive shell to be bash, yes. but better rewrite make file to be understood by POSIX shell.

i, myself, use bash for my regular user

the display manager is a tui written in bash. i want bash to be my shell before? a login

i can't help with that. i never used login managers due to never needed

today I had dhclient sigsegv each time the upstream modem restarted. it received and logged the new client configuration, but then 12 seconds later said "connection closed" and segfaulted without actually setting the IP, leaving the host disconnected. pastebin.com/raw/d31gRZbd

any idea what the deal is? I haven't had this happen before freebsd14. Also are those numbers in brackets supposed to be PIDs? because they keep changing...

dmr104: chsh -s bash ?

that will change the current shell for the current user to bash from whatever you had

i have also found htat as long as you have bash installed.. and hte makefile says uses.. bash, it usually does it magic.. for example tpm (tmux plugin manager) needs bash to do some work but i use zsh

lw, ping

git clone github.com/freebsd/freebsd-src --single-branch -b releng/13.2 --depth 1 # seems exactly the same as if i take out '-b' so is -b redundant?

Title: GitHub - freebsd/freebsd-src: The FreeBSD src tree publish-only repository. Experimenting with 'simple' pull requests....

Is there a way to make a zigbee gateway in FreeBSD? I see a Linux project for it, but I'd rather keep my systems freebsd if I can.

lx: vermaden.wordpress.com/2023/06/28/freebsd-jails-containers

Title: FreeBSD Jails Containers | 𝚟𝚎𝚛𝚖𝚊𝚍𝚎𝚗

lw: i meant to tag you. for dladm mention in the above blog post

Anyone know how to increase pointer sensitivity for Apple Magic Trackpads? Looks like everything else is working, except I want to adjust the sensitivity higher. It uses atp+wsp+bcm5974 tunables seem to be making a different (first thing I tried was scale_factor, to no real avail).

Tried bumping it up to 4, 10, 12, 32 and it didn't get any faster.

Oh, [-1.0,1.0] lol

so the default is 0.0 ?

Seems so.

jbo: hi

meena: hah, i knew i wasn't the only one who remembered it

dmr104, Did you figure out GNU ln -r,--relative option yet? It's actually a very helpful to the human feature. Humans always get very confused by symlinks and having the program do it is helpful.

But it is totally non-portable. Fortunately you can always figure out how to do it without using that option.

dmr104: if the script is written in bash, change the #! line to be "#! /usr/bin/env bash", your own login shell doesn't matter for that. for the Makefile, if the commands really require bash, you can try adding ".SHELL: name=bash" but i'm not sure if that's sufficient

ln -r is not a bash'ism.

ln -r is a GNU ln ism.

First, symlinks are a special file that stores a value. That value could be almost any set of characters. If it matches another file name then the operating system will follow through the symlink to the symbolicly named file.

When someone is making a symlink it is usually best to make it relative from the referenced file. I recommend cd'ing to the target directory first. Then using TAB file name completion to get the value correct. "cd /usr/local/bin && ln -s ../foo/bar/baz ." that will create a relative symlink.

One could also simply "ln -s ../foo/bar/baz /usr/local/bin/" and have exactly the same result too. Same thing. But you can't use TAB command line file completion then because we are not in the target directory. And people often don't understand that the "../foo/bar/baz" part is just a string of characters at that point.

does -r make it create relative symlinks automatically? that sounds useful enough to add to our ln

With the GNU ln -r,--relative option one can do this "ln -s -r /usr/local/foo/bar/baz /usr/local/bin/" and it will do the work of converting the value into a relative symbolic name value.

YES! That's what it does. Very helpful. Never absolutely needed. But humans have had problems with that for years. It's helpful.

crazy winter weather stil -15c snowing i had enough of winter

It's been -20C here the last four days with a dusting of snow. It's cold but the snow is pretty.

Typical places where it might be used on a Debian GNU/Linux system for a few examples. ln -r /etc/apache/sites-availble/local.conf /etc/apache/sites-enabled/ will create ../sites-available/local.conf correctly. Places like that.

dmr104, If the Makefile uses bash then it should be setting "SHELL = /bin/bash" there. Which you can then override on the command line "make SHELL=/usr/local/bin/bash" though I install bash and make a symlink "ln -s ../usr/local/bin/bash /bin/" so that things using bash Just Work without needing to modify them. That feels minimally invasive on FreeBSD to me.

But regardless if a Makefile is using a non-/bin/sh shell then it _should_ set SHELL to indicate that need. That's what it is there for. (Though always using portable shell is better.)

rwp: i checked the make(1) manual page and it doesn't actually mention $SHELL, there's a much more complicated method to change it using .SHELL target. (probably because having make respect $SHELL would break loads of makefiles)

i wonder if $SHELL works anyway

That's probably on me to not know that. I know System V style make very well but honestly I do not know BSD style make anywhere near as proficiently.

i think the easiest answer here (aside from just fixing the makefile to not require bash) would be to make it do 'bash -c "..."' ... which is terrible, but if it works

tbsm is a login manager written in bash. I want to run the bash script /usr/bin/tbsm : should i do this in the rc.conf ?

dmr104: what does the documentation say about how to start it? it looks like you're meant to set it as your login shell?

dmr104, /usr/bin/tbsm? Shouldn't that end up being installed in /usr/local/bin/tbsm? And what is the #! line in that script? The above suggestion "#!/usr/bin/env bash" is probably best, though it pains me to see it, and say it.

ah no, the manual says to start it in .profile. so i'd do that

(github.com/loh-tar/tbsm/blob/master/doc/01_Manual.txt)

Since I have been ranting a little let me say that I mostly use GNU make, but use none of the GNU specific features. I mostly use automake to generate portable Makefiles.

where is the manual? what are you referencing?

the link i just posted is the manual

Looking at that I will suggest again "ln -s ../usr/local/bin/bash /bin/" so that things using bash Just Work without further hacking needed.

oh

manual say there may be better ways than this suggestion

The actual verbatim suggestion in those docs is to use .bash_profile, which is not a bad idea if you are using bash as a command line. I do that in order to keep .profile clear and I customize .bash_profile for bash use.

I conceptual problem i am having is that .profile is read after login prompt, but a login manager is before one

It makes me squirm a little to see the suggestion of putting a display manager in a .profile/.bash_profile though. Does it properly detect that you don't want it when logging in from a remote system over ssh? Just for example.

i wouldn't recommend using an rc script because it appears to have no way to actually authenticate the user, so that would immediately log you in with no password prompt. but if that's what you want, you could try running it from /etc/ttys...

dmr104, Where does it say tbsm is a login manager? I see that it says it is "tbsm is an application or session launcher".

dmr104: i don't think that's right... some display managers log the user in, this one seems to be something you run after logging in

oh dear

rwp: that cold is brutal.. i am not at those temps.. single digits F here, just saw a big birch tree destroy a picnic table.. that was "fun"

look at the screenshot on the webpage: loh-tar.github.io/tbsm - it shows a normal login with the linux getty and then tbsm runs after the login

anyone know any "tricks" to break up git diff into smaller chunks, like file pbased?

Title: tbsm | A pure bash session or application launcher. Inspired by cdm, tdm and krunner

voy4g3r2: undo the commit (git reset HEAD~) then commit it properly. :-)

voy4g3r2, Worse is that a group of us did a Lord of the Rings marathon, all three movies, the extended cut length, at a friends house and their house they kept at 60F/16C and though I knew it was going to be chilly there I did not expect that much cold while inside. I froze myself yesterday all day.

ah well, i wrote the install makefile to /usr/bin as the incorrect location for nothing, What login manager should i use with wayland?

lw: well i am the one that made the git diff and "properly" is what i am trying to figure out

dmr104, This is not a login manager but tbsm is a launcher for different Desktop Environments, making it a session manager. Log into the vt console. Run tbsm. Select which DE you want to start. That's what it is doing. All makes sense.

i have about 40ish man page files i updated "to fix" links and i am like.. probably should make them 1 per file instead of a group.. to make it easier to get accepted

voy4g3r2: well, after you undo the commit, you can just do 'git add path/to/file' then 'git commit' to commit that one file. then repeat that for all the files

rwp: i got it like 70F in the house.. wood stoves rule! but man it is brutal outside..

(git reset HEAD~ will undo the most recent commit, if the commit you want to fix isn't the most recent, it's more complicated)

Wood stoves are pretty awesome! But also not for everyone. It's a best fit thing when it is a best fit.

rwp: well i suppose i could forget about the l/m and just do things that way

I use "git add -p ..." *a lot* and then pick what I want out of the menu for the diffs.

rwp: i hear ya, i have multiple sources of heat for those times power goes out.. heat important.. thankfully we have a stream on property to get "non potable" water for going to bathroom.. and could even use said water and wood stove to "sanitize" water

But there are git managers such as emacs-magit which proctor an interface to do the same thing that many other people like better. (I hate magit because it breaks other things so I can never use it.)

lw: yeah, i just checked out the source, making updates in my local copy.. no commits.. so just a bunch of files

Also I am often using "git rebase -i ..." to organize the changes. Between "git add -p" and "git rebase -i" you can do anything.

voy4g3r2: so you didn't commit the changes? just add/commit each file then. although if you mean a way to automate that... i'm not sure. is 40 PRs the right way to do this? that's a lot

although i suppose 1 PR with 40 commits is okay

lw: i am not sure yet.. it is why i asked the question :)

i do not even know how to do the PR yet, that is on the list

voy4g3r2: btw, if you're on main (or whatever), before you do anything like committing, make a local branch: "git checkout -b my-manpage-fixes". that will save you a log of hassle later

this isn't *technically* required to make a PR but you almost always want to do it this way

voy4g3r2, I am a little nervous at hearing that too. I will commit locally for things, and I will push them to my other machine as an upstream repository for me, just in case my local laptop or desktop dies horribly then I have my work on my other machine. When I am ready to publish that when I push to the main upstream.

i am doing this all in a bastille jail now

before i was just doing on a machine and it was giving me "crap"

i swear this man pages broken link bug WILL be resolved

you should still make a branch even if you're doing it in a jail, this is about making it easier to manage the git repository itself

i got oer 631 potential "issues"

Another hint: If I am doing work and the upstream is active then I might have uncommitted work-in-progress but I want to sync with the latest daily upstream. "git stash push && git pull && git stash pop" will save all uncommitted changes, sync local with upstream, apply local changes again.

if you commit directly to main, then you update main later, it will not be clear which commits are yours and which are from upstream, and it's harder to work with

whereas with a branch you can do something like "git log origin/main.." and it'll show all the commits on your branch

well okay then.. git checkout -b my-manpage-fixes it is

Yaaas, got it figured out. Have to change the acceleration profile to custom values. It's a bit convoluted, but whatever, it worked. Pointer speed via touchpad is faster now.

actually tbsm is referring to a non-bsd directory structure: looks for /usr/share/applications . So I need something like lightdm after all

dmr104: that would be /usr/local/share/applications, sounds like the app could do with a patch to fix hardcoded paths

If, and only if, I am working on main then I will "git rebase main" to move my changes on top of the new main. I always work in the rebase flow and not the merge flow.

After rebasing local changes on top of main's pointer then can "git branch my-wip" to take my changes off the main branch and onto my own branch.

is there any equally good display manager than lightdm? Does lightdm support wayland?

wiki.archlinux.org/title/Wayland#Display_managers says it doesn't support wayland but can launch wayland compositors. never used it myself

Title: Wayland - ArchWiki

i guess that means you need X to run the display manager, which sounds awkward

That sounds about typical for the immaturity of Wayland though.

i have labwc which supports wayland only

check your compositor though, some of them (like sway) explicitly don't support being run from a display manager

being at the end of a cable modem loop is NEVER fun and there must be issues somewhere.. speeds are in the 100~

rwp: to be fair that's more lightdm's fault than wayland's, there are wayland-native display managers like sddm. i think kde's still doesn't do it yet

i take it git clone | git checkout is NOT the path forward?

git clone git.FreeBSD.org/src.git ~/src <--- is what i did to populate /usr/src and then i do a git checkout -b /usr/src ?

Title: src - FreeBSD source tree

voy4g3r2: git clone git.freebsd.org/src.git; cd src; git checkout -b my-branch

Title: src - FreeBSD source tree

that's if you want to create a new branch, because you can't do that in the clone (as far as i know) like you would for an existing branch

Title: Chapter 3. The Working Copy | FreeBSD Documentation Portal

i may need to read this a little more but it takes me down the path i have been taking

I used to use sddm with xorg. Tried to use it with wayland/hyprland, but didn't work and I didn't want to spend the time. I should try again, because I liked sddm when I used it.

voy4g3r2: ok, i understand your original question if this is how you did it. i don't recommend doing it this way, although it's okay for small changes, your changes are much bigger

voy4g3r2: what i would do is this: create the local branch (with checkout -b), make your changes, add/commit each file in whatever way you want (perhaps some commits can have multiple files), then once you're done... make a pull request on github. because that's by far the easiest way for non-committers to manage larges patches and get them applied

when i save files to a flash drive then connect flash drive to another computer it shows user:group of files differently even though i have the same username account on the other computer. why's that?

voy4g3r2: that said, it might be worth waiting for a committer to offer some advice on how to organise these changes, like how many commits you make

is it internally on the flash drive stored as UID or?

alepzi: user/group ids are always stored numerically on a filesystem (except on NTFS which uses SIDs, a sort of GUID-like thing)

so how do we not plug in flash drives to computers and get some random guy having permissions to it, not us, because they have our previous UID now?

if you want to mount a USB disk on a multi-user system and restrict who can access it, i would suggest mkdir -m 700 /mnt/private and mount it there, then only root can access it

if it's formatted with FAT (msdosfs) you could also use -m700 when mounting it, that's specific to msdosfs

when i add a user can i set which uid for it to use?

yes, see pw(8)

ok then i'll just do that for my personal account everywhere then the flash drive permissions will always show my user account

-u

do you know what the uid range can be? like 0-?

65k roughly

alepzi: UID_MAX is 4294967295

whoa that's huge

it's a 32-bit unsigned int on modern platforms, some older platforms use a 16-bit int in which case the max is usually 65535 (or 65534 or -1 is special)

s/or -1/if -1

i sent a patch for a port and ended up becoming its maintainer. this is how they get you. before long you're maintaining half the ports and live in a dungeon dreaming of what the sun looks like. (or so i heard)

alepzi, I don't know if FreeBSD can do a mapped uid mount like Linux systems can do. But on Linux one would mount with a uid map to map uids to your current uid.

This problem is probably why US Flash storage is still mostly FAT32 or NTFS still after all this time because then there are no uids to map.

If all of the systems you are using are your systems then the normal thing is to ensure that your user account data always uses the same uid:gid across the entire collection.

NTFS still has this problem on Windows, which is likely why USB sticks come with exFAT by default nowadays

what's the easiest to set up and/or least impactful way to relay local mail off a new freebsd host to my main mail server?

doug: docs.freebsd.org/en/books/handbook/mail/#dragonFly-mail-agent

Title: Chapter 31. Electronic Mail | FreeBSD Documentation Portal

(if you're on 14.0, dma is in base)

where are the instructions for gdm ? There is no man gdm ?

Do you have gdm installed?

yes

How did you install it? Source? Ports? pkg?

pkg

help.gnome.org/admin/gdm/stable/gdm.html ?

Title: GNOME Display Manager Reference Manual

pkg info -l gdm | grep man

nothing

I think it likely that GNOME simply does not write man pages anymore.

the webpage is old. no mention of wayland

is there any readme from the pkg ?

pkg info -D gdm

Hmm... Reading the docs it looks like FreeBSD uses fuse to mount exFAT file systems. That's interesting. Surprised it's not native though. It's fine though.

How does one format an SD card to be exFAT?

exFAT was patent-emcumbered for a long time which might have something to do with that

Oh! It looks like I needed to install exfat-utils and fusefs-exfat to make this work.

well, all the instructions are for files which i cannot find on my system.

dmr104, GNOME stuff will be designed and documented for files in /etc/gfoo but pkgs by policy will install everything in /usr/local/etc/gfoo so that's the main place to look.

i am running /etc/periodic/weekly/310.locate in an attempt that locate will work with the updated files. has taken 5 minutes so far. I have ports installed too

lw++

Uhm... What's the frequency that you see in that path for creating the locate database? What's it say there?

dma comes on 13.2 as well and was dirt easy to set up

i use postfix because i prefer having the same MTA everywhere, but it's nice we have something in base now that isn't sendmail and works for 99% of users

rwp: i don't know. i am a bit new to bsd

anyone use docker on freebsd?

It says /etc/periodic/weekly/310.locate so it will be run "weekly". You would have to wait a week for the automation to catch up to these new files.

I use locate as generally useful to locate files that I know have been there but I don't remember where they are now.

If I just installed something using "pkg install foo" then I can list the files installed with "pkg info -l foo".

aha /usr/local/etc/gdm is there. But if I trigger the action now, it will update the database immediately

cat dmr104|grep docker

If I have a file and I don't recognize it then I can ask pkg which package installed it "pkg which /usr/local/bin/aatest" returns "/usr/local/bin/aatest was installed by package aalib-1.4.r5_13"

Yes. If you run /etc/periodic/weekly/310.locate now then it will scan the file system right now and everything will be up to date. But that is a pretty heavy action. Might run for a bit. That's why it is only done weekly by default. Often it is configured to be run daily.

i bet you could make locate update much faster (and more frequent) with zfs snapshots and diff

experemental, I am not using Docker on anything. Sorry. No help from me.

that might be a bit too invasive to be the default method though

lw, That would be a cool highly efficient way to generate differential builds of the locate database. :-)

that took longer than expected but thanks rwp and lw .. we got a branch and now i can have 1 main release-14.0 source and do my changes in isolation

grep -v rwp\

voy4g3r2, It's useful to work on a branch, fetch the main branch to sync with upstream, rebase your work branch on top of the new upstream, periodically as needed.

yeah i can see that, because i have a strong feeling they are going to ask for -CURRENT diff at some point

but my experience with getting -CURRENT working has been less than stellar

experemental, I can not read the runes of your writing, it is obviously of an ancient language of which I do not speak. Here in this channel using the common tongue will be more likely to get information and questions answered.

you can write manpages on -current and test them on -stable, just use "man ./manpage.1" or whatever to read it

Are there changes to the -man macros between current and stable? That concept rather shocks me! I would expect those to change once a decade at most.

rwp: i doubt this has changed since the 80s tbh :-d

i have a head a few.. the mandoc sementac change is the big one

so some do \fB or .Xr

there are QUITE a few maybe 12-20 that have links to things that do not even exist anymore

some that are .Xr but are deprecated but good for historical.. so .Eo .Sy .Ec magic

some moved from category 2 to category 3

so you get a mix of some man pages with something(2) and some with something(3)

How many of those changes in the current CURRENT versus STABLE? I expect (without looking) that you could develop on your current system and then do a final check at the very last moment just to be sure on CURRENT.

that is my hypothesis

and figure this bug has been open for 7 years, i do not expect much violatility

rwp: protocol downloaded 100%

voy4g3r2: btw, if you commit your changes to a branch of stable/14, then you want to patch current, you can merge/rebase your 14.0 branch onto current, you'll just have to resolve any conflicts. might be easier to start with -current manpages though

but must now go cook some food

`sysrc sendmail_enable=NONE && sed -i.bak 's/sendmail\/sendmail/dma/g' /etc/mail/mailer.conf && sed -i.bak 's/^#SMARTHOST/SMARTHOST myrelayhost/' /etc/dma/dma.conf`

not sure how that could get any easier

i do not want

(rebase basically takes all the changes you made and tries to apply them to a different branch)

i have different Y issue s

experemental: Can you not spew nonsense here?

making a new zfs volume, anyone ever got the error "g_dev_taste(zvol/zroot/vms/foodisk0) failed to g_attach, error=6"?

it still seems to have run because i can see it in zfs list -t all

alepzi: iirc this can happen due to a race if you have zfs set to not expose zvols to geom, geom notices the new device but then it vanishes before it can attach

it's not fatal because zvols do not depend on geom unless you specifically want to create geom things on them

hm ok, weird

it is while i'm creating a bunch of volumes and snapshotting them in a row

i'm not sure 100% sure about it, but i remember reading something like this recently... i can't remember where