@zygoon@fosstodon.org avatar

zygoon

@zygoon@fosstodon.org

zyga aka zygoon, Dad, FOSS programmer, Canonical, ex-Huawei, ex-Samsung. IRC: zyga, GitHub: zyga, GitLab: zyga-aka-zygoon photograph bees and shitpost about the state of FOSS from time to time.

Snapd developer with unlimited enthusiasm.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

mackuba, to golang
@mackuba@martianbase.net avatar

It's insane that I can have a conversation with this thing and in a few moments and a few back and forths build a working program in a language I know nothing about (I don't know Go)โ€ฆ don't tell me this isn't huge ๐Ÿ˜ณ

You: One modification - this current code counts all matches of all regexps against the text. So if one line of text matches two or more regexps, it will be counted as 2 or more. I want a given line of text to only be counted once if it matches
You: What's the command to compile this to a binary? ChatGPT: To compile a Go program to a binary, you can use the go build command followed by the name of the Go file you want to compile. If you want to specify the output binary name, you can use the -o flag followed by the desired name. Assuming your Go file is named main.go, you can compile it to a binary named my_program like this: You: Looks like
You: Great, that works ๐Ÿ‘ Is there some way to compile this program in "release mode", with a more optimized version or something? ChatGPT: Yes, you can compile your Go program in release mode with optimizations using the go build command with the -ldflags flag set to -s -w, and the -trimpath flag set to remove the build path from the resulting binary. Here's the command to compile your program in release mode: You: I mean, not optimizations for binary size, but optimizations to make it faster? ChatGPT: Ah, I see! In Go, optimization for speed primarily comes from the compiler's default optimizations, and there isn't a direct flag to enable additional optimizations beyond what the compiler already does. However, there are some general principles you can follow to improve the performance of your Go program: Profiling: Use Go's profiling tools (go test -cpuprofile, go tool pprof, etc.) to identify bottlenecks in your code and optimize them. Profiling helps you understand (โ€ฆ)

zygoon,
@zygoon@fosstodon.org avatar

@mackuba but ioutil is deprecated :-)

juliank, to random
@juliank@mastodon.social avatar

How to render dependency graph as a graph, what are the understandable approaches there?

zygoon,
@zygoon@fosstodon.org avatar

@juliank I wanted to do something like that and realized that at the end of the day people handle lists better than graphs. I would render it as a list of direct and indirect dependencies. Graphs of even tiny size quickly render into illegible artwork. Unless you want to invest in a kick-ass layout, rendering and navigation widget, avoid graphs.

zygoon,
@zygoon@fosstodon.org avatar

@matk @juliank the problem with graphs is thatvyou need useful user interface for the worst case. Nail that before touching the easy cases. This is why I think lists are better for communication with humans.

zygoon, to random
@zygoon@fosstodon.org avatar

Are you building your packages after a short break? Did you remember to run "sudo solbuild update"? It may save you some head-scratching and debugging.

Happy packaging ๐Ÿ™‚

juliank, to random
@juliank@mastodon.social avatar

Here is a comparison of the new apt(8) UX vs the classic apt-get(8) one for the install command.

Not duplicating the listing for dependencies means that despite having more padding (of sorts) it actually is comparable in length, but much more readable.

zygoon,
@zygoon@fosstodon.org avatar

@juliank I cannot wait for the "it breaks my workflow" bug that is inevitably coming.

pid_eins, to random
@pid_eins@mastodon.social avatar

This is such a bad bad API compat breakage:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e81cd5a983bb35dabd38ee472cf3fea1c63e0f23

It's used all over the place in userspace. In systemd we use it:

  1. to detect if a block device has partition scanning off or on
  2. In our udev test suite, to validate devices are in order
  3. udev rules use it for some feature checks (in older versions of systemd).

And it's even a frickin documented userspace API:

https://www.kernel.org/doc/html/v5.5/block/capability.html

So much about that nonsensical "we don't break userspace" kernel mantra.

zygoon,
@zygoon@fosstodon.org avatar

@pid_eins There is a bugzilla https://bugzilla.kernel.org/

zygoon, to random
@zygoon@fosstodon.org avatar

The working branch of my userspace parser for binary profiles that are usually parsed by the kernel can now parse any profile I throw at it.

I don't do the DFA unpacking and analysis yet but given that I'm stuck at home (extreme allergy to whatever is in the air recently) I think I will get to that today.

I've been working on improving visibility into what apparmor_parser compiles by effectively de-compiling it back to human-readable data.

The code is up on
https://gitlab.com/zygoon/go-apparmor

zygoon,
@zygoon@fosstodon.org avatar

@roddhjav AppArmor has very little visibility into the userspace-kernel interface. The best you can do is run apparmor_parser with the dump option to see what the parser claims to do. The second problem is lack of up-to-date documentation that's readable and accessible.

The goals of the project are:

  1. Document the userspace-kernel interface in detail
  2. Provide a 2nd implementation (apart from the kernel) that reads binary profiles and displays them in human readable format for debugging
zygoon,
@zygoon@fosstodon.org avatar

@roddhjav apparmor has, sometimes, surprising semantics and as a part of figuring out where the problem lies, the text profile, the included base/abstractions, the compiler (aka parser) or the kernel, one has to figure what the binary profile actually says. Once that is demystified it is clear which side is wrong: userspace or the kernel.

drewdevault, to random
@drewdevault@fosstodon.org avatar

If you're asked to sign a CLA and/or copyright assignment before contributing to a FOSS project, refuse, and make a hard fork of the project instead.

Take direct action for free software.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault I think there is also the part where the copyright holder chose to share it under a free license that is being neglected here. The drive-by contributor did not create the project from scratch. Regardless of what one prefers, it is only fair to recognize this side equally.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault, pit yourself in the shoes of a hypothetical developer who works on something and CHOOSES to make it available under a copy left license, hoping to make it useful to individuals while leaving the gate open for business engagement that wants to avoid copy left.

Accepting a contribution, even relatively trivial, changes that ability.

It is again something that each project and developer needs to come up with a policy for, but sweeping the problem under the rug is not a fair response

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault yes exactly! This is why accepting or not accepting contributions is a choice, just like choosing to release the source under some license.

My whole point is that it is not a clear cut black and white decision, and I can imagine myself making the same unpopular choice in certain circumstances.

A copyleft+CLA project is just one particular way of trying to make non-proprietary, gratis software.

Until this problem is solved, we will keep seeing people try possible solutions.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault, but the community does benefit from the original availability. I understand it is not a popular choice among the foss crowd, but I think we can recognize that maintaining asymmetry is one of the reasons the software is non-private to begin with.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault, how can a business close something that was not entirely owned? It cannot.

It must start with the same owner releasing something significant enough to warrant asymmetric contributions (small) from the wider community.

You can argue that over time, this is immoral, but the basic stance is too strong as the owner had every right to make that decision, and all the participants understood the deal.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault, but then you also say that all the contributors participated in immoral activity, and I think you can not make that claim for them without taking their agency away.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault I think I either don't understand something about your opinion or I can not make my opinion clearly enough.

I wish we lived in a world without CLAs...

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault in your opinion, is that equally morally wrong as making proprietary software or is there a gradient there?

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault right but I think this is related: the alternative to the copyleft+CLA is, in many cases, full proprietary.

I don't like it but I think it's in some ways better than not having the software in the public at all.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault wait, is it?

I don't get it, how can you equate that.

CLA is purely a business model. One of the very few in the FOSS world.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault no, it's really just a business model: someone hires a bunch of people to create some software and then releases that under *GPL + CLA hoping it gets popular and can be sold (either as a service or as license).

There's no moral part there yet, it's just a decision to operate in this specific way.

The moral decision is individual making the contribution. The rules are clear.

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault the contributors are not ignorant and CHOOSE to play along.

They could have forked but didn't.

The non-free angle is entirely up in the air: it's possible, I totally agree on that but it's not guaranteed. Only certain types of software is susceptible to being closed as a SAAS.

GNU requires CLA for entirely different reasons but it's (more or less) clear why they do that.

In both cases the contributor chooses to contribute under the rules set out by the project or fork or not.

zygoon, to random
@zygoon@fosstodon.org avatar

Having lots of RAM in your desktop or laptop is liberating. No need to remember to close one project and focus on something else. Ability to run large test environments. I feel cramped on a 16GB system I use from time to time. My main desktop has 128GB and my main laptop has 96GB and with this anything I'm working on fits perfectly.

I really want the industry to move away from base 8GB models, that's just instant e-waste in many ways. The cost of memory is too low to do that.

drewdevault, to random
@drewdevault@fosstodon.org avatar

Soliciting names for a redis fork (take note of the trademark)

zygoon,
@zygoon@fosstodon.org avatar

@drewdevault ridiculous

fedora, to fedora
@fedora@fosstodon.org avatar

Flock to Fedora 2024 call for proposals open until Apr 21!

If you are interested in speaking at Flock, now's your chance to reach out. We have three tracks (main, CentOS and Friends, and Fedora Mentor Summit), but we also have themes that we're looking to explore.

  • Accessibility
  • Reaching the World
  • Community Sustainability
  • Technology Innovation and Leadership
  • Editions, Spins, and Interests
  • Ecosystem Connections

Apply!

โžก๏ธ https://communityblog.fedoraproject.org/flock-2024-cfp-until-april-21st/

zygoon,
@zygoon@fosstodon.org avatar

@fedora I'd love to go to flock but I think I will pass this one and visit the next one in Europe.

zygoon, to fedora
@zygoon@fosstodon.org avatar

I've posted a whole slew of updates for and that bring snapd up to 2.61.2 - a small point release bringing bug fixes.

https://bodhi.fedoraproject.org/users/zyga

F40 is the last one to go.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • โ€ข
  • provamag3
  • rosin
  • normalnudes
  • everett
  • DreamBathrooms
  • ethstaker
  • magazineikmin
  • thenastyranch
  • Youngstown
  • GTA5RPClips
  • slotface
  • khanakhh
  • vwfavf
  • kavyap
  • megavids
  • mdbf
  • Leos
  • Durango
  • tacticalgear
  • InstantRegret
  • cubers
  • osvaldo12
  • ngwrru68w68
  • anitta
  • tester
  • modclub
  • cisconetworking
  • JUstTest
  • All magazines
    •