Linspire's CNR Goes Multi-Distro 171
S3Indiana writes with news that Linspire is opening its Click 'N Run installation software to other Linux distributions. After 5 years of development on CNR, the new site cnr.com will be a single source repository for Linux users. Distributions to be supported initially during 2007 are (alphabetically): Debian, Fedora, Freespire, Linspire, OpenSUSE, and Ubuntu; other distributions will follow. See the FAQ and the screenshots for more details.
A welcome new contender, but... (Score:3, Insightful)
But *works* should include the following:
- installs new software correctly, in default and custom locations
- uninstalls old software correctly
- updates old to new software correctly
- is aware of and can work with custom-installed libraries and dependencies (i.e. EVERYTHING doesn't have to be installed using this system, some stuff can be compiled from source or downloaded from third party).
- is scriptable through some command-line interface
- isn't a pain in the neck
As far as I know, none of the software installation systems out there for any platform meet all of the above requirements. InstallShield for MS systems probably comes closest, but is definitely not perfect (nor even "good enough" imho). Until something comes out that I consider "good enough", I'll keep hand-rolling, thanks.
$.02 from an old slackware user.
Not what it is, what it isn't. (Score:5, Insightful)
In all seriousness, that's pretty much the crux of it. From TFA: Basically, their problem with apt-get is that the tools are harder to use, and that it's distro specific. Their aim, if I'm understanding it right, is to offer one tool that would be the same across distributions, offer the same software to each, and be extremely easy to use. In short, rather than each distro having its own package management system, they could all use CNR and appear the same to the casual user.
If you use apt-get, you probably aren't going to be interested in CNR, or really anything that Linspire is doing, frankly. But I think there are a lot of people not using Linux right now, and who are confused by the differences between distributions (not to mention the very concept of distributions in general) who would probably be receptive to the idea of a standard packaging/installation system that was distribution-agnostic.
Re:Repositories? (Score:3, Insightful)
Not aimed at us (Score:4, Insightful)
I'm happiest supporting people on Ubuntu/Kubuntu because that's what I run. If I can now also give them *easy* access to the software they know by name, without me having to intervene to do the messing around with wine or whatever, this can only be a good thing.
Finally (Score:5, Insightful)
What is consistently one of the biggest gripes about the Linux desktop? I know one I hear and see often is the difficulty of of installing Linux applications when the disto does not provide them. Autopackage has tried its best to cross the gaps, but even its main programmers concede its hard to do all the cross distro work (that is often cleaning up messes) when there is no financial reward to inspire you. Its not exactly exciting and low hanging fruit like a new 3D snow pluggin for Beryl.
If Linspire does this right then here is the solution for one of the last few big complaints on the Linux desktop- new programs will be easy to install on any distro soon after release. If soon the user does not have to care that they have Ubuntu or Suse when a new Gimp or Crossover Office comes out then the Linux desktop might be ready for a big run. One main problem of course as this is a closed solution to the problem- removing both kinds of free in order to make it happen. Yet users pay for software now on both of the other primary desktop platforms, so I don't think many will care. If this is done through "partnerships" then Linspire might make a large amount of money in this new gatekeeper role while boosting marketshare of the Linux desktop in the hard to get at home market.
Glad this finally happened. Now the last big problem- the lack of drivers- will be fixed the only way it can be: increased marketshare. We hope...
Re:Linux is a failure (Score:2, Insightful)
Re:A welcome new contender, but... (Score:3, Insightful)
Most Linux distros already have package management that does all of the things you mention. This is in fact one of the major benefits of using Linux over Windows. Apt-get (for example) works like a charm, assuming what you want is in the repos (and in Debian, it generally is). I guess you are referring to cutting-edge software, that isn't in the repos yet?
What is interesting about CNR is something completely different than 'software installation that just works'. CNR can provide a standard and easy-to-use way to purchase software, for example, DVD playback, codecs, etc. - things that Linux distros can't legally include by themselves (in some countries at least). While the typical Linux user may not care much about these things, some people do, and in particular some organizations must be 100% legal, no doubts allowed. For them, a convenient way to spend a few dollars for mp3 playback and get it legally may be a big thing.
Re:Repositories? (Score:3, Insightful)
This seems to be the equivalent of Perl's CPAN for Linux, combined with Download.com (for ratings and reviews), with a pretty GUI thrown on the ability to also offer Commercial Software through the same interface.
When Jane and Joe set their new computer, and want to get Firefox (they know its good and all, since they're "techie" friends all rave about it), its nice to go to one place to install (presumably with no hassle or fuss. Likewise when they want to play the DVD of their god-childs play on their computer, they can go to the same place, give their Credit Card info (assuming its not on file I would guess), and download a DVD player, again from the same place. All through a GUI that users can drive.
I'm not sure if their software fulfills on this promise (I haven't personally tried it), but if they can its a HUGE step toward making Linux accessible to the masses.
Re:A welcome new contender, but... (Score:5, Insightful)
The biggest points are:
- You can't specify where to put files
- You can't specify the menu structure (The menu structure matches the CNR store exactly, making it easy to find stuff)
- It's trivial to use.
It MUST NOT have options like a CLI or flexibility.
Currently it is extremely easy to use for non-Linux people, in fact it's easier than getting software onto a PC.
With your requirements, I suggest another piece of software.
Isn't it acceptable that different programs handle different needs for different groups of people? Must you really break something for me so that it works like you want?
Please, just accept it the way it is and say, it my be great for you but not me!
If you see CNR as just another entry into the package management/software instillation, you don't get it--please keep quiet and don't ask them to change software that is perfectly suited for others.
Re:Not what it is, what it isn't. (Score:5, Insightful)
I would consider myself an above-average user, but when I go to install something that's outside the scope of my repositories and get shot down by the dependency failures... that's when I get a little peeved.
If Linux standardised, I'd be sure to recommend it to my friends and family. Even the dumbest "For Dummies" distros aren't simple enough for Joe Bloggers to use.
Re:Enough CNR like things... (Score:5, Insightful)
Only it's one of those unlucky pieces of software that isn't in the repository because of some dumbass nerd licensing pissing contest. So you google around some and find out that some guy is running an unofficial repository that contains it, and you only have to alter a couple of files to include the repository address. And then you can install it! Easy!
Only, half the time some guy's repository's latest version is three months out of date because some guy has a life, so now you're downloading the source and compiling it yourself. But hey, it's all so easy!!
The problem with all these programs is that all they do is introduce middle-men. They intermediate and abstract. They get in the way. John just released Foobar 2.1, which fixes a nasty bug you are dealing with. Only that doesn't help you, because you need it in your Distro repository, where Mike maintains it. Only he's too busy arguing arguing with Jack, Sally, and Javier in the Distro dev mailing list about life, liberty and who's more hardcore about the meaning of Free Software. I'll get the new version in 6 weeks, if I'm lucky and Mike doesn't resign in a hissy fit.
I've been through all this shit as a Linux user, and I got sick of it. Fuck Mike, I like dealing directly with John.
There we go.... (Score:3, Insightful)
While this sounds like aimless Linux zealotry, this will probably be another flurry of people who complain needlessly about Linux package management without having bothered to use it or understanding it superiority to any other sort of package management.
Man, an InstallShield-like installer is a step BACKWARDS for package management! I've had to spend hours and dozens of reboots in Windows getting my software right! It's a task that with aptitude or synaptic gets done in under 20 minutes, no reboot, full use of the machine in the meantime. ISVs should be embracing .deb and .rpm. C'mon, it's not that difficult to learn how to package for 3 or 4 major distros! If you know how to write a Makefile you should know how to package software.
Re:Not what it is, what it isn't. (Score:5, Insightful)
Re:Not what it is, what it isn't. (Score:3, Insightful)
Ditto. This is IMO one of the biggest weakness of Linux, and conversely Mac OS X's single-file
Re:Not what it is, what it isn't. (Score:5, Insightful)
As you quoted from the article:
I hear this argument a lot, and if it's a concern then I guess it makes sense to try and address it, whether it's by trying to improve what's offered, or by trying to educate people about alternative ways to do things. I don't really understand it, though, and to be honest, the huge amount of packages that actually are in a distribution archive (esp. Debian, which I use), is one of the big reasons I prefer to use open source instead of closed source. (I'm sure there's a certain amount of me just being used to it, too.)
I think people often look at things the wrong way when trying to compare Windows with Linux distros, because they work fundamentally differently. The reason there are different installer types is that they're different systems, and the installers and packages are made to match the system. (Granted this doesn't mean it couldn't be improved and made more compatible.) Windows shouldn't be compared with Linux, it should be compared with RedHat, or Ubuntu, or Gentoo, or whatever, because the distributions are what operate at the same level as Windows. The fact that they use similar or identical apps and are often compatible with each other just makes Windows stand out because it doesn't.
I don't see the issues as being as much between Linux and Windows as being between Open Source and Closed Source, because the distribution model is what makes the difference.
Microsoft strongly encourages third parties to release closed source apps, as they do themselves. As a result, Microsoft doesn't have a lot of control over the app or how it interacts with the OS. Microsoft isn't legally allowed to tinker with third party apps and throw them into a big Windows software repository. There are some weak conventions about how applications should interact with the system, but it all comes down to whether the vendor actually implements these and does it correctly. (eg. Install in the Program Files folder, use a particular structure in the registry to store settings, and so on.) Some apps follow the conventions properly, and some don't. It's entirely up to the vendor. If I download and install a typical third party app for Windows, it's not unusual that it might be unstable, fail to take advantage of and integrate nicely with other apps I have on the system, and so on. It's very unlikely that a Windows installer will go and download dependencies for me -- chances are it'll package them inefficiently and often unnecessarily, or it'll tell me to go and find them manually. And if I try to uninstall it, I'm usually relying entirely on the independent vendor's uninstall scripts to properly remove itself. I don't know about other people, but personally I've found that they often leave a lot of residue lying around. (Old folders and files, registry entries, broken links and icons, obsolete dll's, etc.)
OSS distribution maintainers, on the other hand, have every right to tinker with the software before they put it in their repository. I know that if I apt-get install something from Debian, it's likely to work with whatever else I have, because Debian's package maintenance team has made sure that the package strictly adheres to all of Debian's policies. I'm not just getting the app, I'm getting a guarantee that it's been tuned to work nicely on my system. Of course, if I don't want that, I can still download the app indepen
Re:Not what it is, what it isn't. (Score:3, Insightful)
Thanks for the link. I've tried to make MSI's from time to time, although only simple ones, and I'm still figuring it out. The main point I was trying to make, though, was that installers for most Windows apps come directly from a third party vendor. There's a lot of depending on third parties to get the installation scripts right, even though there might be conflicting commercial interests, or just general laziness to provide a good (un)installer.
I've had problems using MSI's in the past. I think it's mostly because the vendor's screwed up, maybe intentially on some occasions. That aside, there also doesn't seem to have been anything built into the package manager to protect the system when a vendor screws up. In Windows' case when it's necessary to rely on packages from third parties, I'd love to have a package manager that can reliably roll back broken packages, as well as whatever changes the application in the package might have made when it was in use, without relying on a rollback script that came with the package. If Microsoft's able to make MSI's more reliable in the future, then that's great and it'll make my life at work quite a lot easier.
Re:Not what it is, what it isn't. (Score:4, Insightful)
Linspire's particular advantage is in believing that "right" is trying to find a way to lure the most users possible to their distro by making it friendly and easy-to-use. Users shouldn't have to worry about dependencies; they shouldn't have to ever even see a message warning them about it, worry about having to resolve it, or worst of all, have a package refuse to install because of "unresolvable conflicts." Solving this involves research, which means time, which means that most people would just as soon spend the money on Windows or OSX and call it good. They work, they rarely tell you that you're missing something, and if they are, they usually provide an easy path towards finding the software you need, which also usually installs without complaint.
Sure, a high-end user may want the option to continue using a particular version of a particular library, and damn the dependencies. That's what "more options" and "advanced" buttons are for. You need to build in user-selectable levels of complexity, rather than having the system dictate terms.
Re:Not what it is, what it isn't. (Score:4, Insightful)
That's a strange way to put it. There's only one OSX. Naturally applications written for OSX will run on OSX. Applications written for Debian Linux will run on Debian Linux, too. But there are many Linuxes, each with different versions of the basic components, and different goals and development teams. The open system works just as well as the closed system if you stick to software that's designed for the particular OS you're using. If anything, the Linuxes are far better at running applications written for other operating systems, than either Win or Mac OS are.
That's not to discount the fact that the lack of a single distribution with Microsoft-level relative dominance is a problem for Linux. But it should be recognized that that's the issue. The choice is mainly one between quasi-monopoly and competition. The Linux situation can be improved, but as long as there are multiple popular distributions, it's never going to be as simple as it is with a tightly-controlled single OS from a single vendor.