Would it help if the removed systemd?
Debian Project Leader (DPL) Jonathan Carter has described the key problems in the Debian community as not a lack of funds, but rather a shortage of volunteer developers. The project is tiny in comparison to the many thousands of organisations that depend on it. Ubuntu is based on Debian, as are other well-known distributions …
Wouldn't go so far as to say remove the well known init system/binary logging platform/umpteen other things but would suggest making alternatives more equal in treatment. Remember that Lennart himself suggested that the well known mega package might not be best for Debian when he was presenting.
The 'row' was a wake up call for many by the way. I personally am init system agnostic as I have never actually had to change any init related defaults on any Unix like system that I have used. However I have moved over to Slackware and OpenBSD over the years.
Yes indeedy. A Debian which included install options for SystemD, openrc and good ol' sysvinit would attract back all those Veteran Unix Admins who do all the compatibility work.
But hey. let's not get too "enjoy the choice" here, that might be a bit too radical for a distro on which so many others are based...
"...I personally am init system agnostic..."
I can't say I've ever been agnostic but I very reluctantly put up with systemd until one day it decided to stop a "start job" that had been running for over a minute -- unfortunately without mentioning that this "start job" was fsck. That totally fscked a three terabyte drive.
I have a few more juicy stories like that, but that was the one that led me to remove systemd from everything that would let me do it without a complete rebuild.
Unluckily there are some utilties which demand systemd, like the Hewlett-Packard printer software which worked fine after the upgrade to Buster but then fell over after an upgrade to *itself* because of not having systemd running. WTF do my printers care about systemd???
Anyway I put systemd back on the print server, and next time we replace the printers HP won't be on the list of candidates.
Got it in one, there is too much dogma.
While I will support with code and cash, I don't feel the need to touch systemd related files, nor do I feel like debugging the same issues worked out previously in shell.
I accept that there is a good amount of value and some actual problems solved by systemd, I do think the problems addressed by systemd would be better backported into the wider Linux ecosystem.
I want to make install a kernel rather than build a dpkg. I do think that the community around debian can be toxic, especially the self-rightous lot that expect to be treated with respect, whilst treating others as beneath contempt.
I now run gentoo with openrc, and it's blissful, I don't need to touch v much other than security updates.
Ultimately, I'm happy with shell, but I understand that others prefer the syntax offered by systemd.
It's fairly small beer in the scheme of things, but hacking on FOOS should be fun, and systemd instance on running things that work perfectly well without it, are not fun.
XInetd powers a few hundred million of turnover for a client of mine, gambling terminal clients.
It's small enough that a cut down version has been audited and verified.
Yes, systemd can do socket activation, but what exactly do I gain here, and so the problem begins.
Buried in your comment is the key takeaway: Debian enables a toxic community. As Leader Carter said "Currently too many people take on too much responsibility because they feel there is no one else who can do so.” That's it exactly. Individuals who do not play nicely with others are allowed to seize control of maintenance by denigrating the contributions of others. Those who prefer to work in a cooperative environment quickly leave, and only the clique remains. The clique then self-destructs because someone has to be the weakest link each round. Eventually the survivor rage-quits for lack of adoring followers, and the WNPP list gains another orphaned package.
Classic example is trying to ban Tovalds, like you're running his kernel, write your own if you don't like his speech. "You're offended? Well so fucking what!" being an arse is not a protected characteristic.
Funny enough, the most "right-on pc wanker speech from a debconf" I heard came from a Brit.
Wondering out loud how someone survived until adulthood given they were too stupid to find food is relatively polite for an angry Finn.
You've hit on the major weakpoint of the volunteer FOSS model. While many volunteers end up being paid to work on their hobby, many do not, and end up pulled between having to make a living and fulfilling their personal commitment to a project. Shortage of time leads to poor progress, and complaints from users who just assume that package maintainers do nothing else but maintain packages.
I'll be charitable and say that it's often the thanklessness of maintaining a package that causes the aggression - any small request runs the risk of being the straw that breaks the camel's back. (That said, yes, there will always be arseholes in every sphere of human activity)
As for what to do with the money: Once it becomes practicable again, why not pay for people to meet up more often in person? It's time we stopped assuming that every programmer is an asocial hermit - it's simply not true, not even for Linux. An easier opportunity to meet contributors and maintainers would also go towards solving the problem of finding new contributors and maintainers.
This post has been deleted by its author
Do the likes of Ubuntu contribute anything back (financially) to Debian? Just wondering. I mean, a lot of OS vendors are profiting from the hard work of very few people. I would hope that, in addition to punting patches and bug-fixes up-stream, they also chuck a few quid, too?
Well, I doubt if having masses more money would actually hurt, with the caveats of the real problems too much money can ensue --- any more than it would hurt any institution or person.
They might get more volunteers if wealthier and considered more stable.
Still, the point about commercial companies which profit from the work throwing a little back is valid.
Same as the much ballyhooed Heartbleed bug that showed according to the usual vermin the infinite dangers of Open Source; whilst having not much effect comparatively. When only a handful of volunteers were able to work on OpenSSL as many multimillion entities profited from their work whilst not giving anything back.
At the moment the project is playing things a little *too* safe. This is probably why so many projects can rely on it and yet it doesn't quite fire up enough passion from developers. Most people want to be rockstars, not maintainers!
A few very rough ideas that might help (some are opinions that you may not agree with. Some I don't fully agree with! ;):
1) Make the core project a little *more* exclusive. Allow the developers to feel proud they are skilled enough to be part of the team rather than just another pair of "punching hands".
2) Some sort of very simple user package repo (similar to AUR, or even openbsd-wip ports). Somewhere that people can work on their packages in a public place that gets them tested and popular. As it stands it is very hard for outsiders to add / maintain / improve source debs.
3) A formal handbook (not a wiki like Arch. More like FreeBSD). I worry that people might feel they are wasting time learning things that aren't the defacto way of doing things. I.e they don't know if this is an undocumented feature, likely to change or not. This will help increase passion and justify time spent with the project.
4) This is controversial but I absolutely think Gnome 3 as the default target has skilled developers running for the hills. They don't want to deal with it, they don't want to support it. They might even suspect the project is running off track because of it.
So yes, I do feel the project does need a little more diversity *of thought* (I don't give a crap about skin colour or genital shape, that isn't going to solve software issues). It needs to become a little more opinionated and try to fire up technical interest rather than just support other projects that suck all the time and efforts from developers.
> 2) Some sort of very simple user package repo (similar to AUR, or even openbsd-wip ports). Somewhere that people can work on their packages in a public place that gets them tested and popular. As it stands it is very hard for outsiders to add / maintain / improve source debs.
> 3) A formal handbook (not a wiki like Arch. More like FreeBSD). I worry that people might feel they are wasting time learning things that aren't the defacto way of doing things.
Strongly agree on these points. I had some spare time and wanted to update a heavily outdated Debian package. I went upstream and fixed a culprit that had stopped previous re-packaging efforts and realised that most of the old package patches were now obsolete. So re-packaging should be fairly easy today and checking out the corresponding Arch package supported this view.
But reading up on Debian packaging was a pain. There are so many obsolete guides and manuals but no authoritative source of how to do things today. Playing with various packaging tools left me baffled - the amount of quirks and warts made me wonder how come these tools have not been abandoned already. It seems like most tools were written by inexperienced developers decades ago and sadly has never improved much since then.
I wanted to start contributing to Debian, but I left with the feeling that the current packaging system is stuck in the past.
Why not contribute to the packaging system then? Ultimately it's only a bunch of bash scripts on top of ar / tar. The existing maintainers have all jumped this hurdle so probably don't quite feel the need to change for change's sake. But a decent guide and perhaps migrating to e.g.Python 3 would improve the situation no end and may well be welcomed. It's also a non-moving target so easy to demonstrate you haven't got something wrong.
"...reading up on Debian packaging was a pain. There are so many obsolete guides and manuals but no authoritative source of how to do things today. Playing with various packaging tools left me baffled - the amount of quirks and warts made me wonder how come these tools have not been abandoned already. It seems like most tools were written by inexperienced developers decades ago and sadly has never improved much since then.
I wanted to start contributing to Debian, but I left with the feeling that the current packaging system is stuck in the past."
This is what needs fixing. If the twit in charge can't see it he needs to be replaced.
Personally I love the gentoo method, it's a bash script that downloads, compiles, installs.
You can also make binary packages for distribution, it's very easy to host your own packages, share existing packages with others.
I wish that was adopted as a cross distro standard, it's very comfortable for anybody from a BSD background.
For the folk wanting a Debian handbook - https://debian-handbook.info (in a website) / apt-get install debian-handbook / pick up a copy from Lulu. Comprehensive well written book translated originally from French. Covers many of the basics - and how to package in various forms. Written by two experienced Debian developers, reviewed/translated by several more.
Most people want to be rockstars, not ...
I think you've nailed it, purposely or not.
To be a maintainer is to be part of the team.
And in a/any team (*), each member has a place and a task to fulfill.
(*) "... group of people who are interdependent with respect to information, resources, knowledge and skills and who seek to combine their efforts to achieve a common goal".
Thompson, Leigh (2008). Making the team: a guide for managers.
The Linux ecosystem is home to literally tens of thousands of highly qualified coders/programmers, a rich pool of talent which could be harnessed towards the same goal and put Linux firmly on the desktop.
If they all want to be able to shout "Look Ma!!! I rolled my own [fill in here]! and be a primadonna for all of 15', that is certainly not going to happen.
If you have escaped Windows for that long then you might not believe it. But yep!
The start menu occasionally gets random "featured" apps that you might "like". The initial install of Windows also has the start menu littered with this crap.
I really do look at my parents using it and feel sorry for them. My dad in particular has been with computers since the early days and really does deserve better than this. Us youngsters that have stood on his shoulders really have let him down.
Tried him with Linux desktop environments but there was too much learning and breakage for him (he is fairly old now!). Tried him with FreeBSD but the desktop environments still require a fair amount of porting work. There really is very little decent solutions for him and people like him. Windows 2000 was the last "real" OS he used.
It might be, I don't understand what they want. I assume they want more "maintainers", but they should be able to hire those so...
In the end they might have to ask themselves who is Debian for. If it's for servers, then start removing packages. If it's for the Desktop, maybe team up with Ubuntu directly.
I leave this for speculative thought...
In 2010, if you downloaded Ubuntu LTS and installed KDE (full), you'd download nearly 7GB. In 2020, you now download ~4GB.
Ubuntu is dependent on Debian as an upstream, has a smaller number of supported packages and many fewer architectures to support - and doesn't have that many independent developers of its own relative to the size of the distribution - but does have users. Every Ubuntu derivative has fewer and fewer developers and a long tail of fewer and fewer users. if you got together the entirety of capable developers for Debian-derived systems - scrapping effort for Ubuntu / Kubuntu/ Mint ... ... - as you went, you might only get another thousand developers.This is not _just_ a Debian problem - it also holds for Fedora and CentOS, for example You don't have to be a whizzkid at packaging software - the website can always do with enthusiastic folk who can translate documentation / text strings to make it better for non-English speaking/reading folk. But it is a do-ocracy rather than a "fix it for me first"-ocracy. [Disclaimer for openness - Debian user and developer for 22 1/2 years and working on CD release team / web site.]
I've long assumed that the role for the distribution teams is to 1) validate that the various packages play nice together at their released versions 2) on an insane number of hardware configurations 3) without blowing up security 4) in a way that's easy for admins to deploy.
Is this fair? If so, the who is managing 2? I've worked for a couple of companies (including AMD) where I was close enough to the relevant teams to appreciate just how mind-numbing that work is.
Who's validating the hardware? The package maintainers are. This is especially hard where you have a machine architecture where some important packages don't build, for example, or for the ARM ecosystem where you have ?? 40 ?? broadly similar but definitely not identical credit card sized computers. It's another good reason that folk ask for information on hardware quite so often / troubleshooters and problem solvers ask people to ask smarter questions / show their workings and prove that they've at least tried to solve problems themselves.
This got me thinking…
Back in the day, much FLOSS software came about because people were working in research labs of some flavour as their day job, and were working on FLOSS either in their spare time, or as some fractional part of the the day job. This allowed things to get done without people having to worry about where their next rent/mortgage payment was coming from (even providing a fall-back if some people, mentioning no names, ended up living in an office in the research lab for a while, when the money did run out).
But then the Linux OS grew and grew and became something that we all depended on and also something that needed more maintenance effort, for each of the distros out of there.
A prominent upstream enterprise Linux vendor solved that problem, for itself, by becoming a fully paid product, and therefore providing an income stream to pay developers.
Debian, on the other hand, is very definitely on the Free side of the fence. This is a good thing, because it makes it available to anyone who wants it, and it is clearly also a big help to all of its child distros which share its virtual DNA.
But are we maybe now getting to the point where managing a whole distro just becomes too big a task to rely on volunteer effort alone? All of those volunteers also have to occupy a lot of their time in some other way to ensure that their rent/mortgage gets paid (and speaking for myself, at least, I seem to have a lot less energy for doing other "brain work" in my spare time than I used to), so quite probably some of those donations do need to be spent on employing those developers, so that they can be fairly recompensed for their labours?
(And it also reminds me that it's about time that I made a small donation to Debian. If even 10% of users donated even £5 a year, that would surely amount to a fairly sizable sum. And it hopefully goes without saying that any companies that make use of it should donate rather more than that.)
"the foul abominations which are Flatpak and Snap"
And AppImage. The choices for installing the latest InkScape are AppImage and Flatpak.
Yes, I get the idea of providing your own versions of libraries to be independent of the distro* but .... I don't know how the other two are implemented (although snap needs its own daemon!) AppImage AIUI runs in its own compressed filesystem. Just why? A whole extra layer of overhead for what?
If you want or need to do that just stick the thing in its own directory in /opt along with the libraries and put them first in the path.
* But will you be as prompt as the distro in updating them for bug-fixes?
First you need users.
Ubuntu originally got a lot of mindshare by being Debian with a bit of extra polish, easier for ordinary users to install and use, and supported by a benevolent dictator. For a long time, things looked good.
But nowadays Ubuntu sadly seems to be getting distracted with things like snaps, which many people have good reasons for disliking.
And so some people start to consider Debian again, which has also become more polished itself. But before any distro can have developers, it needs users.
I've started to move to Debian again, but in some ways, it still doesn't make things easy. Yes, being a developer should have skills barriers to entry, but even as a potential user, it was until quite recently rather hard to find an install ISO image on the Debian website (you are trapped in a maze of sort of download links not especially alike, few of which point to an actual ISO image), and although the minimal net install image is a clever thing, it may or may not work on your particular PC hardware, and the main install image isn't really all that much bigger (and the proportion of people for whom the difference in download size and speed of anything under 1 GB is surely rapidly falling fast, all around the world (not that it's not a good thing to keep in mind)). By all means also offer jigdo and BitTorrent, but those are really for more advanced users, and everyone has to start at the beginning at some point, so why not make it easier to at least get started?
In addition to the points above, I'd just like to mention that Debian is still the only distribution that I know of that provides a good chunk of its repositories as downloadable isos.
Once you manage to find them, a download of DVD1 and DVD2 of the stable Debian release gives you a shedload of software installable offline. Install from DVD1 then boot your new installation, copy the isos to your hard drive, and mount them both. Register them with Synaptic/Apt and you can take a slow boat to China or disappear up the Limpopo secure in the knowledge that you can install software.
The project also provides update isos periodically as each new point release is defined. There is even a blueray iso with the lot on it.
I maintained Debian servers for a very long time and now I work with Centos. It's only now that I really begin to appreciate Debian as a distro and I hope they will soon overcome the difficulties and attract more contributors, because it really is the best "overall" operating system. I tried contributing in the past, but it takes so much time to submit to even the smallest issue that I couldn't afford to spend all my free time doing so. Perhaps I should take another look at it.
Are the maintainers of major branches/components of Debian employed full-time? If not, perhaps they should look into hiring a few of them to maintain the stability of organisational structure of the project because if those people leave and there's nobody to take over then we're headed towards a disaster.
The biggest irony is the fear of socialism as an ideology not realising the entire open-source software community is socialist by definition. These people spend their personal time working on a product free of charge for the better of the community, while businesses use the result of their hard work to build services and products giving back absolutely f-all. It's sad that only a few big names in the industry actually contribute either financially or in terms of bugfixes. Where's the rest of them?
Imagine if every business utilising Debian contributed $5 every month on Patreon to Debian how the project would like today...
Try to work to get the work done more efficiently. As a developer it's quite hard to create packages and support those packages. Repack them. And redistribute them across different distro versions. It's a Maintenance nightmare. A long as the basics aren't solved, your problem will never be solved.
“100,000 packages is a serious reality on the horizon,”
there are too many split packages in debian, for instance installing boost will pull in a truckload of packages. in arch linux, one needs to install only binaries (boost) and libraries (boost-libs) packages. no wonder debian needs more maintainers.
100,000 packages? And you’ll realistically use what, maybe 5000 tops? Knowingly, you probably use less than 100.
Desktop Linux has needed a standardised container format cross-distro compatible for software for some time and this yet again shows this. Yes there are lots of choices available. But no developer wants to have to repackage a program a dozen times in each format.
I’m no MS advocate but standard installer routines that work on multiple platforms are a good thing. one of the few areas where ms has the edge on Linux land.
Biting the hand that feeds IT © 1998–2020