Codeberg.org

https://social.anoxinon.de/@Codeberg

We are a non-profit, community-led organization that wants to help free and open source projects prosper. We provide Git hosting and other services, such as Weblate, Woodpecker CI and Pages.

(We try our best to be responsive, but we receive many notifications on here. Please consider using another communication channel if you need to share something urgent, thanks!)

Hosted at: https://codeberg.org
Home to: free software! 💙
Blog: https://blog.codeberg.org
Documentation: https://docs.codeberg.org

Public Key

npub1fh736ae0xzk98s59wggazdpqg0g78mg2rwk7qf87ew96cqhc4lrst3ee83

NIP-05 Address

codeberg_at_social.anoxinon.de@momostr.pink

Profile Code

nprofile1qqsymlgawuhnptznc2zhyyw3xssy850ra59pht0qynlvhzavqtu2l3crsdt3k

Author Public Key

npub1fh736ae0xzk98s59wggazdpqg0g78mg2rwk7qf87ew96cqhc4lrst3ee83

Show more details

Last Notes

2024-07-09 21:11:02

Our #CommunitySpotlight presents: @npub1z45…5g0a, a simple self-hosted web app that lets you manage bookmarks for articles you want to read. Use the UI to easily group and search your articles, and even transfer content to your #ebook reader for relaxed reading (if you have one).
Just get your browser extension ready, and save interesting content for later.
Find #Readeck on #Codeberg:
https://codeberg.org/readeck/readeck
Let us know what you think!

2024-07-09 21:05:32

- reply

We don't manage the networking.

2024-07-09 19:56:07

- reply

It's probably not bad intent, but rather crawling without caring. The consequences of wild west's AI gold rush.

2024-07-09 19:31:48

Codeberg is currently suffering from massive crawling traffic from some Huaweicloud machines. We are mass-blocking IP ranges again.

2024-07-06 13:14:27

Does your #FreeSoftware project need some help?
Share a link to your #GoodFirstIssue or related labels for others to check out!
An example for #Forgejo: https://codeberg.org/forgejo/forgejo/issues/?labels=222666&state=open

2024-07-04 22:27:20

- reply

Talking of network effects, you might want to check out https://garrit.xyz/posts/2021-01-23-signal-to-noise
@npub1ws3…halv

2024-07-04 22:26:30

- reply

Apologies for the bad greeting :)

2024-07-02 19:01:45

- reply

We'll think about it. Not sure if there are legal requirements with regards to publishing IP addresses. It might not only contain big (and publicly known) ranges but also smaller abusers like misconfigured CLIs.
@npub1nq3…duqa

2024-06-30 13:58:29

- reply

👀

2024-06-27 19:57:59

Ever tried a #TilingWindowManager for an efficient workflow? Using #Wayland already?
Check out https://codeberg.org/river/river #riverWM, a tiling Wayland compositor.
It supports custom "layout generators" and we found an active ecosystem of tools and plugins for the project.
If you know some #Ziglang, you can start hacking right away, or dive into the #ziglings excercises to learn it: https://codeberg.org/ziglings/exercises
Let us know what you think if you tried it!
#CommunitySpotlight #tilingWM

2024-06-26 23:19:22

- reply

This mug has been Codeberg-approved.

2024-06-25 22:44:27

- reply

Would it help if repository names would support more characters, while profile and org names would not?
This might be actually worth a consideration.
In recent years, there have been a lot of attacks by using lookalike unicode characters, so although technically feasible, many fear to accept non-ASCII characters.
But since repositories of the same author already induce a certain trust, it might be worth taking that one step.
Happy to start the conversation about this in#Forgejo.
~f

2024-06-24 22:16:31

- reply

Retry in a moment. I suppose Mastodon instances are currently fetching the link for previews.
This is a long-term problem AFAICT, see news.itsfoss.com/mastodon-link-problem/ and https://fosstodon.org/@drewdevault/111980869898128673 for examples

2024-06-24 22:02:46

We are performing another round of user interviews for #Forgejo from Wednesday to Friday: If you are a user of #Codeberg or any other Forgejo instance (including self-hosted), consider participating: https://cloud.splvs.net/apps/calendar/appointment/cHr5MB2PoZDp

2024-06-21 00:04:34

- reply

For the record, the first years, Codeberg was pretty much hosted reliably on a single VPS for up to 10k users and it didn't require much attention.
The storage at scale and the Git operations performance (high random IOPS and managing all these small Git files, for example) and database performance are the main challenges infrastructure-wise at the moment.

2024-06-21 00:01:54

- reply

Codeberg is overall stable because #Forgejo makes hosting quite easy. We are discovering some scaling issues (and fixed many in the past), but you probably don't have to worry about it if you don't plan to go beyond 10k users.
Running a public service means you have to deal with all kinds of abuse and growth pain over and over again. 2024 was not exactly a lucky year in this regard.
~f

2024-06-20 21:17:39

- reply

(Of course, rate limiting things by ourselves is still something that, as far as I know, can be worked on.)

2024-06-20 21:16:19

- reply

If you try to distinguish between robots and humans like that, you're bound to block, say, the Tor users. Therefore, "retroactive" is the way to go.
Think of DRM: If you use Netflix in Google Chrome on Linux, you get a considerably worse quality for the same price because you can record your own screen and make a copy. But not on the Netflix app for Windows – it's inconvenient to some, but meant to protect something. Their shows still land on piracy sites within hours anyway. ~n

2024-06-20 21:12:45

- reply

They are not viable in our use case because they would either get in the way of developers trying to get work done, because they raise privacy concerns for the users relying on our platform.
We are also, well, out there, because we are a bunch of optimists that believe that we can get some things done if we act beyond the monoculture imposed by a monopoly. Relying on a platform in a similar market position to provide service would, well, probably be deeply ironic to some.

2024-06-20 12:40:38

- reply

Hi, this is true. Although we do opt-out of AI scrapers using our robots.txt, it is still the Internet and relying on the platform for such protection is not 100% safe, no matter if you rely on GitHub or our platform. We recently had to block IP addresses that were relentlessly scraping our platform but _after_ they caused performance problems. ~n
P.S. WIRED published something on AI "content aggregators" ignoring robots.txt very recently: https://www.wired.com/story/perplexity-is-a-bullshit-machine/

2024-06-20 11:24:12

- reply

It was an outage with our database server. We are still investigating to find the root cause.

2024-06-19 13:40:16

- reply

Maybe check out https://docs.codeberg.org/getting-started/what-is-codeberg/#alternatives-to-codeberg

2024-06-18 21:03:04

- reply

Can you show one of the account managers an example of a PDF file? I am intrigued. ^-^ ~n

2024-06-13 17:39:22

- reply

OK, so it seems like the data is not properly persisted in a cookie during the OAuth workflow. Do you frequently use OAuth in other context? Does it work there?

2024-06-13 17:20:09

- reply

There have been updates in the Forgejo issue. Can you tell us if you use container environments or other similar privacy-protecting measures in your web browsers (see the issue for explanation)?

2024-06-12 21:46:22

- reply

We have support for manual merges. The detection is still not perfect, but works in most cases, and you can always provide the commit id that merged a PR instead of pressing the merge button if it failed for some reason.
The option is "Enable autodetect manual merge (Note: In some special cases, misjudgments can occur)" in the Pull requests repo settings.
@npub1cm0…enqe

2024-06-11 07:03:29

- reply

but the GitHub™ Desktop™ part definitely is ;D
But you're right, I have heard this from people before and the lack of viable – as far as I'm aware, may be wrong – alternatives that let you interact with multiple forges is kind of a problem. ^n

2024-06-11 06:29:20

- reply

What are you currently using, if you don't mind me asking? ^n

2024-06-10 17:18:26

- reply

#Forgejo (and #Gitea, too) likely have a UI that is easy to get started with. We often hear that people are struggling to learn GitLab, although it is probably a powerful tool for teams that work with it daily.

2024-06-09 13:20:35

- reply

Tech sovereignty is not the problem. Using it to isolate yourself Free Software projects is the problem.
You can achieve tech sovereignty by supporting international collaboration, and granting everyone the freedom to use it for their goals without being dependent on megacorps.

2024-06-08 14:21:40

EU citizens, your vote for the European Parliament election matters much more than you might think.
From #ChatControl and #EUGoingDark all the way down to #ClimateCrisis: Some important decisions are due.
We noticed populist right-wing forces co-opting "open-source" by "supporting its adoption".
However, they stand for isolationism (hidden behind "tech sovereignty"), and against international collaboration, openness and transparency; which are what makes #FreeSoftware awesome.
Choose wisely!

2024-06-08 14:04:42

- reply

If you're looking for the quick way out, consider taking a look at https://codeberg.org/Codeberg-CI/examples (although I'd love to hear about how it works out for you) ^n0toose

2024-06-08 14:03:18

- reply

You're right. Have to say, I'm not a CI professional and don't know if the default is for the workflow to get triggered, like, always, regardless of the context or the event that took place – it is not what I would personally assume, although that also makes sense but this seems to be what the documentation seems to be insinuating.
Is there a link to the workflow? Did that previously work? I could ask some other community members about it that know more, the docs may need fixing. ^n0toose

2024-06-08 13:37:47

- reply

That should depend on your workflow? We are not familiar of any other similar reports (of push events not triggering workflows) at the moment.
Example:
```yaml
when:
branch:
exclude: pages
event: [push, manual]
```
(A link could help, but I might have to take a look at it later if I do not straight up forget about it, sorry! 💙)

2024-06-07 23:52:04

- reply

Pretty recent, but #codey is starting something like this: https://codey.ch/

2024-06-07 23:50:31

- reply

I think this is what you are looking for: https://forgejo.org/docs/latest/user/issue-pull-request-templates/#possible-options
(disabling blank issues)
CC @npub1mcn…q6kr and @npub1fnh…0x6h

2024-06-06 18:42:20

- reply

Most self-hostable software is actually not aware of email bounces. Most tools are fire-and-forget (SMTP only). They are not aware of bounces.
We could implement it into Forgjeo, but this likely affects so much more.

2024-06-06 15:31:46

- reply

the idea to report servers sending email to non-existent generic addresses is probably a good thing.
The problem is if you use that server to register somewhere, then remove that inbox 🙄

2024-06-06 15:30:33

- reply

what do you mean by the "cross-reference" part? I'm afraid I don't quite understand.
~f

2024-06-06 11:09:08

- reply

And: If you must operate a server that auto-reports incoming messages as spam, consider **not** registering with such an email address on public servers.
Use a honeypot that no legit provider has any reason to email to, please!
Thank you.

2024-06-06 11:08:01

- reply

Blocklists are not only community-maintained resources, but often charge for monitoring and removal services.
There is an imbalance between small and large operators, and the fact that many people suggested us to just use service from $$company$$ instead of self-hosting emails indicates the terrible situation.
We have written about our position in detail previously: https://blog.codeberg.org/how-blocklists-prevent-the-internet-to-be-decentralized-and-safe.html

2024-06-06 11:05:53

- reply

In the responses to this thread, many users started to blame the IT admin who automatically put our IP address to the blocklist.
Please note that we don't want to blame a single party. Reporting spam and reacting to abuse reports is a good thing.
However, we are concerned about the fact that automation based on accusations from single parties escalates to regular headache for network operators.
This is also far from the utopy of an Internet, where everyone can easily host their own services.

2024-06-06 08:35:10

- reply

The issue is resolved with the help of our provider. Thanks for this and all your patience.
Most of the email was successfully delivered, only few remains in the queue.

2024-06-05 21:42:52

- reply

This is the full story: https://social.anoxinon.de/@Codeberg/112564295466537970

2024-06-05 19:31:11

- reply

We don't want to rely on third parties to deal with confidential parts of our user data, but we should've done better to prevent this situation.
@npub16hk…k9st @npub1tcj…kcla

2024-06-05 17:08:32

- reply

We're still waiting for a response from the abuse department about a potential unlock. We apologize for the inconvenience caused.
We have blogged about the problem when providers like @npub13y5…dlhd (in our case) rely on blocklists as the single source of truth, without second thought or verification.
This caused much trouble and headache for us, but the person who put us on the list due to a config error will likely never know it happened, after all.
https://blog.codeberg.org/how-blocklists-prevent-the-internet-to-be-decentralized-and-safe.html

2024-06-05 16:55:58

- reply

We can't contact them, because their mailbox is unavailable. And sending further messages would probably put us on the IP blocklist again.

2024-06-05 16:55:25

- reply

Please see https://social.anoxinon.de/@Codeberg/112564120658911757
But thank you for your report, it triggered our initial research into why our email server is unavailable.

2024-06-05 16:53:53

- reply

They're okay. To be fair, we took a little long to read the abuse message.
But yeah, they could've investigated first.

2024-06-05 13:40:40

- reply

Okay, the whole story, according to our research:
In Dec 2023, an " IT professional" joined Codeberg with a valid email address. In the meantime, they apparently changed something with their server, because the email address is no longer available.
They are also a very active contributor to abuseipdb.com, reporting every server which sends mail to nonexistent email addresses on their personal server to.
This report triggered the abuse department of our ISP to take down our server.
Thank you!

2024-06-05 12:56:13

Hi guys, our email server was locked by the provider due to "abuse" it seems. We're working on it, please be patient.
Email delivery is currently unavailable.

2024-06-04 08:48:41

- reply

Maybe the real pronunciation is the friends we made along the way.

2024-06-03 20:37:24

- reply

*Keanu Reeves voice* "YOU'RE BREATH-TAKING!"

2024-06-03 14:17:29

- reply

2024 is a weird year for the healthy Internet, it seems.

2024-06-03 14:16:34

- reply

There is rate limiting in place, and there is robots.txt. If a crawler peaks Google, Microsoft and the other big players, it typically means they do more than they are allowed to.
You don't need to ask questions in this case.

2024-06-03 13:18:05

Hi everyone! New person managing the social media account here – many of us were at the #GPN22 in Karlsruhe and, this time around, there were many that were interested (or even involved in!) alternative forges for themselves or their hackspaces and, well, Codeberg?!
I even talked to a person in a train that later turned out to be a Forgejo contributor (the person next to them was a Codeberg e.V. member). We all met up!
So long Karlsruhe, and thanks for all the fish! 💙
📸 (credit in alt text)
https://social.anoxinon.de/system/media_attachments/files/112/552/864/062/583/867/original/9ed8e1fa14c9c730.jpg
https://social.anoxinon.de/system/media_attachments/files/112/552/864/121/145/523/original/095dbcfde3a06107.jpg
https://social.anoxinon.de/system/media_attachments/files/112/552/864/086/585/108/original/3462b500fbaabdda.jpg

2024-06-03 12:50:30

We investigated performance degradation due to excessive crawling of Codeberg.
We often trace back such things to company websites that proudly list their key investors - but don't actually explain what they are doing (or why they need to crawl Codeberg more excessively than the crawlers of the major search engines combined).
It's probably not a use case we believe in anyway, so we blocked several large IP ranges today.

2024-06-02 14:12:57

- reply

https://codeberg.org/forgejo/forgejo/pulls/1680 is actively work in progress, and part of it was split into smaller chunks and merged already.

2024-06-01 21:35:32

- reply

We are investigating.

2024-06-01 21:33:46

- reply

The roadmap in Forgejo is determined by user's needs. We don't plan to offer the same features unless proven they are needed by a significant amount of users.
Forgejo does active user interviews to gather feedback for the development, see https://codeberg.org/forgejo/user-research/

2024-06-01 17:17:28

- reply

We can confirm - we once had issues with *new outgoing IPv4 connections* for a while due to network issues. Technically, users didn't notice. Practically, CI was completely unusable, because all projects rely on Github in one way or another.

2024-06-01 17:11:19

- reply

this would be rather hard to maintain and keep up to date.
Some features are also implemented in both forges, but slightly different where the differenes might not matter too much.
One example is repo search using Git grep that was initially added to Forgejo in February 2024 after some months of work https://codeberg.org/forgejo/forgejo/pulls/1594, and was added to Gitea in March 2024 https://github.com/go-gitea/gitea/pull/29998
@npub174k…c9yt

2024-05-30 11:26:45

- reply

https://forgejo.org/compare-to-gitea/ provides a good read. You can also check the initial announcement here https://blog.codeberg.org/codeberg-launches-forgejo.html but more things have happened since.
@npub174k…c9yt

2024-05-10 11:40:54

Anyone considering how to break the #StackOverflow #monopoly already? Any #federated alternative work in progress?

2024-04-22 20:05:35

- reply

Found it. https://codeberg.org/forgejo/forgejo/issues/3375

2024-04-22 12:08:39

- reply

can you give us an exact timestamp so that we can check logfiles for the problem?