We built Steiger after getting frustrated with Skaffold's performance in our Bazel-heavy polyglot monorepo. The main pain points were:
The TAR bottleneck: Skaffold forces Bazel to export OCI images as TAR files, then imports them back into Docker. This is slow and wasteful
Cache invalidation: Skaffold's custom caching layer often conflicts with the sophisticated caching that build systems like Bazel and Nix already provide.
Ha, neat! I’m building a dynamic programming language in Rust called atom (https://github.com/dmeijboom/atom). It’s an interpreted language with a bytecode compiler. Will definitely check it out.
I don’t get the appeal of Traefik. If you want an easy to use reverse proxy that works well, pick nginx. Want something simple for self-hosting? Take a look as caddy. For Kubernetes, try out Envoy Gateway.
We’ve been using Zalando’s Postgres Operator before in production and we recently switched to cloudnative-pg. We didn’t experience any issues so far and I’m a big fan of their design choices (where you have a single user, single database for each micro-service that requires one).
I used zalando and surprisingly hit a bunch of config issues where the backups won't run on an s3 endpoint that isn't aws or wal-e now wont run because it's got no db credentials but does the seaweed endpoint.
now my one db-one user-one pod "cluster" is dead because it won't elect itself leader. I simply cannot kick it correctly to revive it.
This happened a while ago and was related to the trademark/copyright issue. It’s an immature move and I can’t take this seriously. All this does is creating unnecessary drama surrounding the Rust foundation.
Many years ago, I read a comment on Slashdot saying "the fact of a programming language having a "code of conduct" makes it ripe for everlasting drama and politics."
The reason slashdot is no longer relevant as the go-to tech discussion site is just because everyone who loves hating on "modern" things like CoC stayed, and everyone else left. These days it would be more strange to see a major project NOT have a CoC than seeing one. There is drama and politics in any large org whether or not there is a CoC or not.
Slashdot was sold to another media company which had other priorities. Are you aware of its "redesign" drama strictly followed by that same new leadership? Many users left then and there.
I left a long time ago. I think around the same time as CmdrTaco. But the place was basically just tech Luddites and Microsoft bashing so it became a bit repetitive. Also: HN.
Slashdot is a good example how a vastly unpopular redesign to make the site more "modern" can bring the whole business down. It's not even the bad redesign per se; it's sticking to it, and not reversing on it. A real shame in my opinion as I think that auto-moderation system is the best I've seen.
It was ridiculously easy to game the Slashdot moderation system. You knew what topics would get upvoted -- positive about Linux, negative about something like SCO during the lawsuit period. It was not difficult to accumulate karma.
I'm not sure if that was the fault of moderation system. This is more of an echo chamber effect which exists everywhere where there are humans involved. Not sure if technology can change that.
When I read about codes of conduct like this I worry about humnan relations. Does remote work cause the need for CoC? The over prescription of interactions seem like sci-fi to me. I haven’t worked in an office since the last millennium.
No, CoCs became a trend way before covid made remote work possible for the masses. They initially were demanded by project members after high-profile scandals of some kind, usually racist or otherwise discriminatory bullshit, sexual assault on conferences or bullying (Linus Torvalds for example used to be infamous for his language), and then other projects (or their members) wanted ones as well as a preventive measure for the future.
For what it's worth, I dislike CoCs because many of them are written with the assumption that people will behave like utter trash without being explicitly told not to and I grew up with the old "Don't be an asshole" rule [1]... but given what happened in the past where people were clearly incapable of not being assholes and communities having splintered over it, they seem to be inevitable.
I dislike CoCs because they aren't what they claim to be: Instruments to protect people. They are tools of power, ready to be used when needed to bully or oust someone. I therefore really don't understand the broad acceptance or even call for Code of Conducts.
I think this is a reasonable concern. The most recent episode of drama out of Rust is a pretty clear example of problematic behavior by some leadership person, and so it'll be interesting to see whether equally clear consequences will result. In other words, it'll be interesting to see whether their code of conduct is worth the bytes it is stored in.
> Instruments to protect people. They are tools of power, ready to be used when needed to bully or oust someone.
They are both.
A CoC, like other weapons, is more or less value-free. The CoC itself doesn't care what you point it at. It's a question of who is wielding it, and who they're wielding it against.
The problem is most people will in fact not be assholes, but the ones who have a disproportionate effect, because there's no direct penalty. Someone can't stand in your yard and loudly shout personal abuse for very long without getting hauled away by the police*, but there's no such enforcement mechanism online.
[*] yes, it can be more complicated than this. you get my meaning.
> The reason slashdot is no longer relevant as the go-to tech discussion site is just because everyone who loves hating on "modern" things like CoC stayed, and everyone else left.
That was certainly not my experience; I left because all discussions were monopolised by a few seemingly unemployed accounts to shame any and everyone into agreeing with whatever the vocal minority at the time was pushing.
When I stopped reading it, the only people left were those who were constantly using shaming language on anyone who didn't agree about things like a code of conduct, or words are violence too, etc.
The fascist part of it is forcing people to use them and threatening them when they don't understand the need to address a non-problem with terminology, duties, and responsibilities they don't agree with.
You can find no trace of childish bickering in the language, tools, libraries, or documentation. I'm not sure why people doing silly things off in the distance would bias you against using Rust, the use of which would not involve those people ever.
While I think your statement is true, a few reasons:
1 - I will inevitably end up on some mailing list or forums when researching libraries, correct way to do things, etc, where such people interact.
2 - As 'governors' of the language, how can I trust them to make sound investments in the future? I don't want to just learn a language and tooling as a one off.
3 - I don't want to support organizations that act this way, even if the language were perfect.
1) This is an accurate complaint. By my estimation (on the official discord) they interact in such a way for an amortized three of the 1440 minutes in the day.
2) The language is pretty great. Further serious investment is going to be primarily in the third-party libraries department.
3) You need make no contributions, financial or otherwise. I am probably a net drain on the foundation's resources.
> I don't use Rust, but the constant childish bickering has turned me away from even caring to.
Same. I want to use it because technically it's the right choice for a few projects, but when I talk to people in my organization I still label it as "too immature and volatile to adopt today, but worth watching". I was hoping the adoption by Linux and Microsoft would be the forcing factor to stabilize it, but the governance aspect still seems chaotic. I'll keep watching and hopefully it calms down such that adopting it won't be a risk. Or maybe the good parts will just get adopted by a more stable (w.r.t. governance and standardization) language.
Rust is pretty stable and clearly past the point of being immature and volatile. A few issues that the internet has blown up lately doesn't change that.
This, I think, captures the problem quite well; "please don't do this, all it does is create drama!"
Maybe so, but the best way to end up with well-intentioned people absolutely hating each other is to never allow the airing of grievances.
I'm hugely in favour of kind, empathic, and thoughtful communication, and I don't like the "people yelling at each other"-type of projects to the point I simply don't participate. But part of that is also accepting that sometimes people express themselves in less-than-ideal ways, and allowing for the fact that sometimes conflicts happen, and that some amount of yelling and drama is necessarily for a healthy community (or indeed, most relationships between people) because we're all flawed and emotional beings. Rust took a good idea and pushed it to the extreme, which rarely works out well.
Seems to have worked well enough for OpenBSD, OpenOffice (and then LibreOffice from that), Jenkins, OpenSSH, Apache Server, WordPress, Inkscape, Webkit and Xorg.
When did Java get forked? The closest case I can think of is the Oracle/Google lawsuit, which 1. Oracle lost, and 2. wouldn't have even been a thing if Google had forked rather than reimplementing.
I don't see how one could describe this as immature. Immature would seem to be just complaining endlessly, memeing, or otherwise being immature. Instead, they're unhappy with the direction of Mozilla/Rust, and found a perfectly good solution to the problem. That's awesome. This seems like one of the many benefits of truly open source software. You can fork not only to change code, but also culture.
Its actually not. The most recent drama was caused by the Rust Project, not the Rust Foundation, which are two completely separate entities. This seems to be a common misunderstanding which is being spread around here, including the top comment to this post.
The ongoing drama emerging from the leadership / foundation causes real damage to the language. I'm not sure if I like this move of forking Rust, but something has to change. And the change must be as visible as the drama to help Rust regain the lost reputation.
I recently started implementing the Postgres protocol in Rust (https://github.com/dmeijboom/postgres-conn). So I guess I’ll be experimenting with creating a Postgres proxy which translates PRQL on-the-fly.
The TAR bottleneck: Skaffold forces Bazel to export OCI images as TAR files, then imports them back into Docker. This is slow and wasteful
Cache invalidation: Skaffold's custom caching layer often conflicts with the sophisticated caching that build systems like Bazel and Nix already provide.
Currently supports:
Docker BuildKit: Uses docker-container driver, manages builder instances
Bazel: Direct OCI layout consumption, skips TAR export entirely
Nix: Works with flake outputs that produce OCI images
Ko: Native Go container builds
Still early days - we're planning file watching for dev mode and (basic) Helm deployment just landed!