The Random Thread

Charlie_ said:
In the end, mat(portmgr) revert it.
After all, Yuri continued to make multiple commits even after being asked to revert the change.
Click to expand...
Sounds like an unfortunate situation, but without knowing what the communication even looked like, let alone the technical fallout - I'd be very careful to NOT pass any political judgement.

No, I don't follow the development of FreeBSD or the Ports Collection closely enough to make informed conclusions about the actions of an individual dev. It's not out of question that the chain of understanding or trust got broken somewhere along the way. A lot of other possibilities are not out of question, either.

I'm just hoping this gets resolved with level-headed approaches before it snowballs into a debacle where it's hard to separate facts from uninformed commentary offered by outsiders.
 
In the end, mat and des handled this matter.
Yuri didn't even leave a comment, but immediately after mat reverted the change,
Yuri made several commits and subsequently modified multiple ports.
Thirty minutes after Des reported that it was still broken even after the revert,
Yuri made another commit without addressing the issue.
 
You might be a little surprised when updating ports using Git.
Code: 
Sorry, this was inadvertent.
Fixing it.

Yuri

On 7/20/26 21:53, Andre Albsmeier wrote:
> Was this intentional? Just wondering as I got a 90 MB ctm delta 😉
 
Charlie_ said:
It appears that updates to the ports tree have stopped.

296947 – cad/opencascade: Version 8.0.0 broke cad/py-gmsh due to the undefined C++ type std::in_place_t

bugs.freebsd.org bugs.freebsd.org
Code: 
because the repo is frozen?
Click to expand...
Possibly ports repo is in surgery to omit inproperly committed binary file?

Cannot find any announcements in FreeBSD-announce, FreeBSD-ops-announce and FreeBSD-git MLs (looked for ML archives, as I'm not subscribing with them), though. And not in FreeBSD-ports ML I'm subscribing, too.

Not sure such a work is actually possible or not, nor it's the actual reason for freezing (currently) 10 hours or not. Possibly announced to committers only, but I'm not a committer.
 
I'd like to raise the possibility of an Internet routing disruption somewhere between the committer and the machine actually hosting the repo. Reason being: unrelated websites (like old.reddit.com) were accessible to some of us (like me and ShelLuser ), but not to Espionage724 ... and we're on different corners of this round planet. It's basically not out of question that severe weather somewhere has damaged the cables leading to a major Internet node.

A DDOS on the datacenter or a power outage in the area (or even a moron who doesn't realize the impact of messing around with gateways) are also possibilities.
 
Ports git repo seems to be (somehow) rolled back and redoing commits excluding offensive ones.

A commit message mail sent to committer-only ML was CC'ed to me for commit 18745d525008d4d76a79bbf36ffae42bd9066e8d at Thu, 23 Jul 2026 02:00:52 +0000.

This should be the redone one corresponding to commit 0e18fa41b4ccedbb3b82abc4f0d3557b8d569f5e which I'm the author.
If all process are done in-order, 44 commits are remaining after this one.

Currently not sure how it goes on next pull, but at worst, disposing local repo and freshly cloning again could be requested. Let's wait for next announcement!

Note that, if I understand correctly, the repo we (non-committers) can access is read-only mirror of the master repo that committers work on.
So the email CC'ed to me would be from the master repo, as cgit.freebsd.org doesn't show the commit.
 
cracauer@ said:
The ports repo was affected by the same thing that blew up most releases' `pkg update`. Had to be repaired.
Click to expand...
This was actually a separate incident that had unfortunate timing. The cluster builders were attempting to enable the repo-side bits needed for pkg-rwhich, but a bug was discovered that blew up the wrong metadata. It was missed in testing due to the size difference in orders of magnitude between the cluster repo and test repos.
 
T-Aoki said:
The PR seems to be restricted.
Click to expand...

Dropping a note that it was restricted by the author in case it was a sensitive issue. There's nothing bad in there, but I'm going to summarize it instead of open it because I think that's in poor taste- the reporter had noticed that they weren't allowed to push as a result of the freeze and assumed it might have been a security incident, so they erred on the side of caution. It was filed against our infrastructure component.

We sent out two notices within the first day internally because committers were most impacted, then sent out the public one once it became clear that the outage would persist for a little longer.
 
I tried downgrading my laptop's BIOS even further and bricked it 😅 (turns out USB BIOS recovery on Dell is bs and ironically let me flash something incompatible)

CoD MW2 (2009) runs on 15.1 with Wine 11.0 :D but seems unstable with WineD3D (32-bit Vulkan doesn't work for DXVK; iirc something with Mesa and DRI=on?)

Screenshot_2026-07-25_20-08-14.png
 
You must log in or register to reply here.
Back
Top