Hi all! After upgrading to pkg-2.4.x I get the following error when running: cd /usr/ports/packages/All && pkg create -a && /usr/sbin/pkg repo /usr/ports/packages/ /<secure>/repo.key

pkg: /usr/ports/packages/All/pkg-2.4.2_1.pkg is not a valid package: no manifest found

sgs: there's a related thread on ports@ which might provide some clues ("pkg-2.4.1 broken?")

I would like to know if there is a way to request an app to be ported to FreeBSD. Who would I submit this request to if there is a way to do it?

Which app is it?

zen-browser

I ran across it in Manjaro Linux. Can't remember if it was from the main repos or the AUR.

... of Arch.

Hmm... "# Re-do process to pick up [possibly] redefined $rc_conf_files" Under what circumstances could there be a difference between the first time and an additional time?

And in the new proposal there is no delay in that very tight loop. If I saw that code I would think it was a coding error for certain.

rwp: what are you referring to? An open review or commit?

rwp: why would you put a delay in rc.conf? that would make every rc script slow for no reason?

bdrewery: referring to reviews.freebsd.org/D53597 i think, which someone posted earlier

Hmm... I guess that was in the scrollback from yesterday. I just got back from a trip and looked at IRC and saw: < ketas> oh loooook reviews.freebsd.org/D53597

there was no delay before

Right. Hence my question under what circumstances could there be a difference between the first time and an additional time?

if rc_conf_files has new files

And how might that happen?

that's the whole point of that code

when files loaded redefine it

Okay. I see it now. It's initially this: rc_conf_files="/etc/rc.conf /etc/rc.conf.local"

But then someone redefines that to be somethign else in /etc/rc.conf such as rc_conf_files="/etc/rc.conf /etc/rc.conf.local /etc/myotherlocalrc.conf"

yes

And so it runs through and loads files not yet loaded, again, to pick up that new myotherlocalrc.conf file.

yes

periodic.conf needs it actually but it looked weird to me at first and then kevans pushed me to rewrite even the place i got idea from

And then the proposed change would allow myotherlocalrc.conf file to extend that further to be rc_conf_files="/etc/rc.conf /etc/rc.conf.local /etc/myotherlocalrc.conf /etc/andanotherrc.conf"

yes

up to enough times

ketas: this is the sort of thing you should explain in a commit message, fwiw

yeah probably

there's review for when this was added too

That's actually the answer to my question. In case it is redefined. But then I ask... What should the behavior be if someone redefines it as rc_conf_files="/etc/onlythisfileplease.conf" ??

and vendor.conf

put them there as refs?

or just one i mean

unsure how to explain more eh

can always improve

In that case it is too late to source onlythisfileplease.conf because it has already sourced the other files. Kind'a a weird case.

unsure what happens then

you can't undo it eh :p

or i mean

I am sure that the redefine that removes previously defined files is ignored. Because they are already sourced by that time.

in the end you can break init

well lets see

only that file gets loaded?

and then, unsure

anyway

It's keeping track of files that are already sourced in $sourced_files and only sources files that are not listed there. But by the time it is redefined the previous files have already been sourced.

the original code doubling has better explanation in it

i could search it up and add as reference

somewhere

some 2017 or when it was eh

so what do you propose?

it can't unload them

Yes. Now that I understand what it is doing the comment "# Re-do process to pick up [possibly] redefined $rc_conf_files" makes sense to me now. I should have noticed it as the clue I needed to know that but it slipped by me anyway.

it could load defaults again but

Honestly I think I don't like the feature at all, even in the original one time case, because it adds a level of complexity that I don't think is actually needed.

Does anyone need multiple levels of /redefining/ the list of rc.conf files? There are already three locations available.

maybe

I try to avoid being too parochial and avoid saying that because I don't have a use that others won't have a use. But it seems that three locations is already many.

could use vendor.conf and rc.conf

And also rc.conf.local

hmm wrong one but

even more exotic things

The new code would help me if it had a comment that hinted that the local admin might redefine rc_conf_files in each of the new rc_conf_files up to the 25 loop count limit times.

Being difficult to describe in a comment is a clue that the feature is subtle and has subtle interactions.

The description mentions "...new loop that runs as long as needed or 25 times..." but it does not define what "needed" means there.

that was supposed to mean until files stop appearing

:p

But files are not "appearing". That is why I was asking, how was this working? Because I assumed some network mount or something where files were being created asynchronously or something!

the original periodic issue is bugs.freebsd.org/bugzilla/show_bug.cgi?id=283427

I have to wait for it to make sure I am not a bot, nice anime image, ... ah, there it is...

on some page that waifu took 82s

unsure what they did there

It's anti-abuse mitigation due to the endless AI scrapers that are taking down every web site. It's the new normal. The AI scrapers and ddos'ers are why we can't have nice things.

the 82s tho

Thanks for the previous bug references. It is the top of the hour and I must run to a meeting now. I have exhausted my free time to look at this right now. Gotta run!

maybe those beagleboard hw design files are extra precious

i can't find the original which added that

that's not it either

:p

Fall 2025 FreeBSD Vendor Summit - Day 1

funnily in the phabricator and bugzilla there are not (yet) used and declined changes too, which sometimes give new good ideas

that's the commit part of it

maybe it just was that

oh yes

Add comment to explain functionality of code

in next commit

anyway

i don't know what real answer is

if it's not slower, it's fine i guess

i don't know what loops or test -r cost

the less the better :]

)

ketas: aren't you tuned to the Vendor Summit yet ?

no

now i looked

it's a break now

yeah

had to confirm location and tz

it's a morning on the other side of planet

why is it round

why it's a ball

nah i closed that

i haven't watched or visited any fbsd even yet eh

perhaps you have to give it a try

Panel: Emerging Security Trends and Their Impact on FreeBSD

this tho...!

an AI generated panel?

LLMs discussing