How big is the file? Dropbox can produce public download links.

If it's under 2 GB, a free dropbox account will suffice. Proton Drive will allow up to 5 GB if you do a few things (making a public download link is one of those things).

TommyC: nimaje suggested 0x0.st which worked fine.

TommyC: 0x0.st/Po2U.zip if you want to try it out. This is for 14.3

Never heard of that 0x0.st, will have to check it out.

to share here too what I shared in #supertux > hm, that binary depends on libfmt.so.11, while I have libfmt.so.12, probably because I use the main branch of the ports tree and not the quarterly branch

I thought we link things statically.

nimaje: Can you have libfmt 11 installed in addition to 12?

I will try with LD_LIBMAP=libfmt.so.11=libfmt.so.12 in a moment, but I have to install physfs first

ah okay

works with that LD_LIBMAP until I try to enter a level, then I get Undefined symbol "_ZN3fmt3v117vformatENS0_17basic_string_viewIcEENS0_17basic_format_argsINS0_7contextEEE", probably a ABI diffrence in libfmt.so.11 and libfmt.so.12

maybe

But at least it runs.

nimaje: What branch do most people use? Also, does that mean that whenever freebsd updates, SuperTux will fail to run?

I guess we need to link libfmt statically

you can't really ship binaries for freebsd that depend on ports, it just doesn't work. ports has frequent or rolling updates so it's only a matter of when it'll break, not if

Alright.

ivy: So how do we get the correct version of libfmt bundled?

I would guess most users use the supertux package so you only need to build your on binary for testing

yeah, assuming this is open source software the right solution is to just have it in ports, so people can install it the normal way. if you do really want to ship a binary... iirc, fmt lets you embed it into your projet fairly easily with cmake? so that would be one option

Well, this is mostly for nightly builds.

we do also have std::format on all supported branches, in case that lets you avoid fmt entirely

(compiler updates are backported, we're on llvm 19 right now)

well, there is more than fmt that can/will break, but for nightlies I guess just using quarterly should be ok

alright

and do we really still need supertux 0.1.3 in ports? hm, probably I should make a patch moving games/supertux2 to games/supertux and put that in a PR

nimaje: I mean, it's milestone 1

Drastically different from milestone 2

nimaje: for a video game, maybe it makes sense to just have 1/2/etc ports

since the old version doesn't become obsolete in the way an old application usually does

pkubaj_, hi! Can I update sysutils/cpuid to new version 20250513? etallen.com/cpuid.html

Patch is trivial. Tested on live system (build and run) and in poudriere 14.3 amd64.

VVD: i already have it committed, just need to push

VVD: ok, pushed

would you guys say docker has finally gotten close, reached or even surpassed jails? i know it's a convoluted question with a possible convoluted answer

but if it has a simple answer i definitely would like to hear it

or in this case read it

people who actually hear their own voices while reading can still say they're hearing it though. Nuances

Jails and containers are apples & oranges. For the part they overlap, I'd say I consider jails to be better and easier. For the part they don't there is no point comparing.

gt: Alver: jails are apples, containers are some kind of mutated fruit that developed several arms and legs and some sentience

gt: but containers do actually run fairly easy and well on FreeBSD

but for actual service isolation that runs persistently, i use jails

jails are containers tho, just generally used to run an entire os environment vs a single process; though doing either is possible. running an entire os environment in docker is frowned upon from my understanding.

Docker containers are too ephemeral for serious business, and adding persistence takes special effort.

pkubaj_, thanks!

using bluetooth tethering right now... it's really showing me why mosh is so important

i really wish there were more tools for a terminal that were tui based just in case i'm ever stuck on mars with low bandwidth

mosh is great

it's very handy, I use it daily

greatly appreciate that the installation process is basically just "have existing SSH access, and install the mosh binary on client and server"

hello. I have a rhel server i want to convert to a bhyve virtual machine. It has a 1TB physical disk, but is only using 16GB of space right now. I don't have enough free space on the freebsd server to make a full 1TB disk image. Is there a way to generate a sparse image from the disk?

(I also have a windows desktop that will need the same treatment, but I'll do that later. I never look forward to windows :( )

gp5st: find a p2v tool that does sparse. dd can to it with the option sparse.

rtprio, oh, i didn't realize that was there. thanks!

not sure if you're going to dd live from rhel or slap the disk in an enclosure and dd it, but both dd's should have it

there probably are a couple of dedicated tools to do it which might have more features

I was going to dd it with the system off. Thinking of doing it on a live system gives me stomach pains. (I actually have 0 reasons, but it _feels_ wrong. Maybe I'm just too old.)

if you're bringing the whole thing down you might think about surgically resizing the disk as part of the migratoin

hmm, that's not a bad idea. The users of this aren't going to store a ton of stuff on here.

ah, they set this up as xfs which doesn't support shrinking

gp5st, When I have needed to do this I have created a new image of the smaller size that I could support but large enough to hold all of the data, cloning the partition table with a smaller size file system, then copied the previous system to the new using rsync with -a --acls --xattrs --hard-links.

I am not a bhyve expert, which is why I lurk here to learn more, but have done that with KVM systems. I would be anxious about using a sparse file system because it has the potential to balloon in size during normal system operation.

but it wouldn't balloon more than the specified size

rwp: rsync or dump/restore that is more what i meant with surgically resizing, yep. not shrinking the partition

I like xfs, it is a good file system, but it is not so flexible with resizing. Increasing the size it is good to increase in multiples of the original size. But decreasing is not supported.

For your MS-Windows task that you mentioned though it is possible to resize smaller the file system. That path is well tested and supported. Shrink it and then binary copy the raw file system. I have done that recently to create a VM and it works well.

For rsync I should mention that -H hard-link detection and copy might need HUGE memory resources. Sometimes it's necessary to "sneak up on it" by copying without hard links being maintained and then fixing those up on a second pass across the data. It all depends upon the system. Just a heads up that it might be a problem though.

In order to copy hardlinks it is necessary to keep a mapping of all files inode numbers with link count greater than 1 (indicating that they are hard linked somewhere) and then duplicating that on the destination. If there are a lot of hardlinked files, such as might happen in a backup situation where the backup uses hardlinks to deduplicate, then this can use a lot of memory.

Thank you rwp and rtprio

is something wrong with opengist in FreeBSD 15.0-RELEASE? I installed the port and it works manually, but the rc.d service doesn't start with default settings (service opengist start -> starting opengist -> service opengist status -> opengist is not running)

swee: did you enable the service too? or onestart?

I enabled it with sysrc yeah

then i would suspect there's some config error preventing it from starting; not every app starts with default settings

yep, from what I'm seeing it looks like I have to do some extra steps, opengist is getting a panic according to /var/log/daemon.log