Linux Foundation unwraps distro normalizer 4.0 The Linux Foundation - the non-profit consortium that gives Linus Torvalds his paycheck and facilitates the growth of Linux and Linux standards - has announced the first beta of the Linux Standard Base 4.0. There are many Linux variants, and they use different kernel releases and software libraries, depending on the technical …
and that was the line, that sort of put this project in its niche.
I don't know, it is not hard to port applications across distros, and some of the distros don't use dependency checking which is the big problem ever since the term RPM dependency hell came into being.
Nowadays there are hardly any problems, sure it happens but the packaging systems all have ways to deal with it.
So, perhaps they are a little late to the party, it is a bit like FHS sure people look at it and see if it offers something, but if they deviate from logic or pragmatism people ignore it.
Good luck to them, but I don't think their certificate will mean much at all, and I think it appeals more to those who cannot already build their own systems. For example do they address the issues of centralized build server, creating different versions for different architectures or for different user requirements? Sloting is useful, and I would love to see that concept further extended, but I doubt they would be able to address that.
And some of the better distros use a rolling release schedule, I wouldn't like to see things held up because people are having to adhere to these standards, and sometimes you want a different init system and to use different baseline command line tools, if they mandate some things as having to be available it causes problems and conflicts, so distros ignore it.
The toolchain for a from source distro is also very complex, it is always amusing upgrading the thing that will then upgrade everything else, if they try and step in there it will cause problems, it is inevitable.
And think about the number of different package management tools available, each being able to determine how the system is laid out, each determining the dependencies, and default configs. There are tons, most of them niched and doing certain things. So, if I am building a lite weight distro I don't want something that brings in a sql database as required, or perhaps I don't python to run the package management system because it forces C++ in gcc. So, I wouldn't like this endeavour to reduce choice, and it has that potential.
My suspicion is people will give this project a nod, but on the whole ignore it, a bit like distrowatch :)
One day, maybe they'll get as far as the POSIX folks did with APIs and the Open Group did with APIs and more. Maybe they'll even get as far as a "best of breed" distribution like the Open Software Foundation did with OSF/1? All this with only maybe a dozen players to worry about.
What you on about, I hear the new boys say?
To which I say, "I rest my case, your honour; the evidence is there for all to see. Too much fragmentation is not only bad for NTFS, history shows that it will be bad for Linux too."
http://www.levenez.com/unix/
Anyway, Suse [1] rocks my world, and the rest of your distros just suck :) :) :)
[1] Insert locally favored distro as appropriate
man's fragmentation.
Daily I use systems of ArchLinux, FreeBSD, OpenSolaris, Gentoo, LFS, OpenWRT, OpenBSD, (and diminishing numbers of Ubuntu installs). it is not a problem, sometimes I wonder about setting up aliases for the different mechanisms, but really it takes all of a few seconds to adapt to each one.
And I get to see the differences, which is useful, it shows some of the capabilities a software package may have.
See, unix is about learning, and continually learning and experimenting, choice and freedom is what drives unix. Sure some want something a lot more static, but there are distros that favour that mindset Madriva, SUSE, RedHat. And I am sure some of those will be more pro this, but others won't.
LSB isn't working, I have no guarantee of compatibility, and nowhere near enough portability to build reliable applications.
Why can't the LSB folks go talk to like MySQL or Java that have built a process that ensures compatibility. It's like we have the Unix wars all over again, what runs on Ubuntu wont on Red Hat, what runs on Red Hat won't on SuSe.
We need people really committed to this, not so focused on commercial opportunity.
We have dropped Linux support as there are simply too many distros, too many possible variations with each distro, too many headaches and not enough bandwidth to adequately test every plausible combination.
Windows may be shit - but at least it's pretty consistent shit.
The little fanboi feifdoms are killing what should be the basis for a good OS.
It is not your job to make sure your package runs on any distribution. Your job is done once it actually builds, with no warnings, on your development system.
It is the job of the individual distributions' package maintenance teams to adapt packages to run on their distribution. If you can help them do that job (perhaps by using GNU autoconf properly), well, that's all fine and dandy; but getting a package to work with Ubuntu is ultimately something for Ubuntu to deal with.
It's the *job* of the likes of Ubuntu to be bothered to make it work. And, as I believe I have already pointed out, if you use the appropriate build automation tools properly then their job (indeed, the job of anyone else building your project) becomes that much easier.
If a piece of software is good enough for the distributions to want it, they will do the work. And if they won't do the work, maybe it's because the software isn't good enough?
Choice may be good in terms of the number of applications you can have diong essentially the same job, but for an OS its a complete PITA. This is where Windows I'm afraid to say and OS X without a doubt, win the day, for most people, they're easy to use. I'm not saying better techincally or whatever, nor am I saying its what you get used to.
A computer is something most people either do work on or play games on or do creative stuff on. To achive that, if they install an application on Windows or Mac, it installs (99.9% of the time) it doesn't whinge that this or that or the other dependancy is missing or whatever because they have x or y distro and not z.
Choose one solution and stick to it, thats update method, GUI, install procedure,whatever. Get all these highly talented people who code for Linux to get the OS foundations as good as possible and give your everyday Windows user a product they can use as easily as they use Windows. Mac users have already largely got what we want, even if we do pay more for it!
Only then will we get a viable alternative desktop OS to Windows or OS X for that matter.
Trouble is, I think a lot of people would rather keep Linux for "Linux" users.
Why didn't your company drop support for all but X? A lot of companies out there will support, say, Redhat and Ubuntu but nothing else etc. That won't help everyone, but you can cover a majority in a scenario like that. Plus Unix admins are smart folks - it's usually not to hard for them to take instructions for one distro and apply them to another.
Leave the AC alone -- he's full of shit.
If his "product" really existed, it was probably Caged. Which means that (a) it won't be missed, and (b) someone, somewhere will have created a Free version which was actually an improvement. And if not, then that's most probably because whatever it did was better done manually anyway.
More than likely, though, he's just full of shit. ACs usually are.
"If his "product" really existed, it was probably Caged. Which means that [...] someone, somewhere will have created a Free version which was actually an improvement."
Reality check: Microsoft's continued existence depends almost entirely on your supposition being untrue. (The same is true for the rest of the closed source software industry. Last time I looked, it hadn't disappeared either.)
On a decent Unix - i.e. one that doesn't enforce the GPL, it's possible to use both free and commercial software, and the followers don't follow the blinkered 'free' view of the world that is frankly just as bad as everything being commercial.
Go get a reality check that there is always a free (as in speech) alternative software for every commercial offering. Tell you what, let's go and pick an easy one : go and find the freely distributable source code for a fully 3D accelerated Nvidia graphics card driver. I'll wait..
One way to standardize something is to force everyone to use a specific thing.
The better way to standardize something is to help create the common communication methods which allow compatibility and "integration" between programs while still allowing for freedom in program selection.
That's the point of ABIs/APIs, and if the LSB focused more on pushing out the "frameworks", the systems, the standards which allow this modularity between components of the OS, and less time trying to make specific software that is the most popular be a "standard" and instead let users decide by natural popular opinions, their job would be much easier.
For instance, keep helping to push a good packaging format which can be easily adopted by package managers so that all the most popular ones at least will adopt it, and Linux users will FINALLY have cross-distro package solutions. If it's a good standard, it will be adopted. Worried about Linux users having a "core base" of programs to work with? Why? Not that anyone can easily install any Linux software, it's a simple install away from whatever repository source.
I like Debian/Ubuntu's way of doing things. If you try to run a program that isn't installed yet, it tells you the package it's in and the command to run to install it.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
