"stuck in the late 90s" in what sense? Building them? Distributing them? (Dependencies?)

I would say the process on getting "maintainer" status.

That they work reliably?

They work reliably with a lot of unseen work. Debian is a great distro as a user. The .deb packaging tools I found to be a huge hassle and made packaging up missing or updated or custom software a pain in the ass.

Exactly, this is why it takes literally years of human effort to get a new version released

Have you built a .deb with debuild and debhelpers? Can you recall all the goofy rules about what needs to go in each file and what each debhelper does? How do you keep all of it in version control and build downstream packages that depend on the changed one? Have you managed to accidently sneak a local build machine dependency or piece of information by accident into the final .deb only to realize it much later?

I found it be a big ball of easily forgettable twine of rules tools and oddities myself that every time I needed to redo something was a hassle. With many foot guns that lead to odd package issues.

Completely agree with your assessment of the building process. The majority of the complexity involved in the process is not inherent

Better being in the 90s than using snappacks or whatever.

Yes but at least I don’t see ads on my Operating System for Viagra. I’ll take 90s boring linux over windoze everyday of the week.

Deb packages just need reinventing by some younger people.

I’m thinking: implemented in nodejs with a cli with emojis and animations.

The GitHub read me should have a minimum of 80 emojis and a meme or two.

The core dependency of the cli should be a 6 month old framework with 92 commits. When that library reaches 1 year old, it should be swapped out for something newer as the sole maintainer will have left their job and/or got bored with the project having spent so long on it

Excellent straw man takedown

> the .deb ecosystem still feels like its stuck in the late 90s linux era to me.

Can you elaborate, please? What do you mean by this?

Quirky, clunky tooling with a bunch of arbitrary shit you have to memorize to use correctly. People enjoy that kind of thing after they learn it.

Enjoy? No. But yeah, it gets easier once you've done it a few times.

Not really enjoy doing, but enjoy knowing arcane shit even if it doesn't make sense.

Have you anything to point? I don’t remember any non-sense in packaging a deb.

I'd like you to explain this a bit deeper. As a user from the '90s I am quite happy with it, but I might be stuck in my ways, who knows? So not saying you're wrong, but what are we missing exactly?

what deb and rpm and any similar packaging are missing is true version control of the whole packaging ecosystem. at present it is difficult to track which combination of package versions has been tested, and you can't easily roll back to a tested combination. the current systems assume that versioning is linear and that a newer version of any package is always better than an older version. downgrading any package so that every user of the distribution can benefit from the downgrade is difficult and confusing.

there are distributions that provide rollback. but as far as i can tell they require you to keep the old version to roll back to, stored on your computer. you can't roll back otherwise.

i really want this to work like revision systems for code, where i can just checkout any old version that was committed.

another feature that i would like to see everywhere is stickiness of the packaging source. currently, if i include additional repos they override the main repo, such that always the newest version is picked from any repo.

this makes it difficult to include less trusted 3rd party repos. i would like to be able to add 3rd party repos such that only the packages that i explicitly install from that repo will also be updated from that repo, while any other packages in that repo will be ignored unless no other repo has them.

in debian it is possible to set priorities for different repos, but that is not easy to manage. the priorities to have each package update stick to the original repo should be default.

guix and nix do provide some of this as far as i can tell, but i am not a fan of keeping every package self contained with massive link trees. (i may change my mind on that some time maybe, but that's what i feel for now)

conary was/(is?) a packaging system that did have both of these features, although, according to some of the developers the repository was a bit clunky and could have been better. but that was under the hood, not noticeable to users and packagers. i loved working with it and i wish foresight, the distribution using it had become more popular so that it would have had the manpower to keep going.

With snapshot.debian.org you can roll back to any version of any package that has been saved. Using that you can even bisect where issues started.

https://manpages.debian.org/testing/devscripts/debbisect.1.e... https://wiki.debian.org/BisectDebian

You can use pinning to achieve what you want with 3rd-party repos, using the origin feature.

https://wiki.debian.org/AptConfiguration#apt_preferences_.28...

A few of my gripes:

* creating Debian packages is a difficult process to learn. There’s a bunch of tools and wrappers around those tools to handle inadequacies and it’s not clear how to do it “right”. For example, if I just have a binary and want to put it into a Debian package without pulling in some random shell scripts that aren’t part of Debian proper, it’s not immediately clear how to do that.

* Non-atomic and imperative install/remove/etc. hooks make it difficult to robustly handle failures. Since there is no file system manifest or sandboxing, there’s no real guarantees that removing a package will actually remove it.

* there’s no spec for a .deb package. This means the only way to correctly generate a Debian package is through the difficult tooling. I understand why this choice is made, and it has a lot of benefits for the continued evolution of the debian project. However, it makes it difficult to build better tooling around deb packages.

The spec for .deb packages is in the deb(5) manual page:

https://manpages.debian.org/bookworm/dpkg-dev/deb.5.en.html

That's not a 'spec' - it's a vague description of the archive names.

It and the other manual pages linked from it contain everything you need to make a .deb from scratch.

I can’t find it right now, but the Debian docs explicitly indicate that you shouldn’t do that because the internal format is not stable. It has changed significantly in the past, and the only “supported” way to do it is with the official Debian tools

The .deb format doesn't change significantly at all, the only changes in recent years have been changes to default compression choices (old ones are retained forever though) and some additional fields in the control file.

The only significant future change I can think of is the mtree stuff, which would add an additional metadata file to .deb files.

There are several warts in the .deb file format that the dpkg folks aren't fixing because of compatibility concerns.

https://wiki.debian.org/Teams/Dpkg/RoadMap https://wiki.debian.org/Teams/Dpkg/Spec/MetadataTracking https://wiki.debian.org/Teams/Dpkg/TimeTravelFixes

If you "just have a binary" then you aren't compiling from source, which means you can't make modifications to that binary easily, which Debian wants to be able to do to fix security issues and other bugs.

There is a large range of documentation, everything from the simple to the more advanced to different packaging niches. The large amount of docs makes that harder to navigate though, and a lot of stuff is moving towards automated packaging these days anyway.

https://wiki.debian.org/AutomaticPackagingTools

Having to use these automatic packaging tools is exactly my gripe. I don’t want to have to rely on whatever scripts someone conjured up if it’s not supported by the distro itself.

I agree it always makes sense to have sources involved for packages that are getting distributed as part of the distro, but what if I want to just deploy my product on a server without the sources?

The tools I am talking about are integrated into, supported by, and used by packages within, the distro itself. I'm not talking about external things that make low quality packages.

If you want to make external packages (such as those that don't come with source), then you can just use `dpkg-deb --build` or one of its many wrappers to shove the binary into a .deb.