When Lycoris released a new version of its popular Linux operating system, Desktop/LX 1.4, this week, it also debuted a new application integration infrastructure, nicknamed AI2 (A-I-squared).The company claims that AI2 “bridges the gap” between commercial Linux applications and the Linux desktop. Already, one major Linux application vendor, NeTraverse, has had its flagship desktop product, Win4Lin Workstation, integrated with Lycoris’ Desktop/LX operating system using the AI2 system. But is AI2 really as revolutionary as Lycoris claims? Even if it is, does it matter?
Lycoris adds new application integration layer to Linux
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
🙂
Why? IMHO it’s balanced and fair. In a nutshell, I argue that AI2 is revolutionary, but is in danger of being ignored because Lycoris doesn’t want to port it.
“Traditionally, installing and working with commercial off-the-shelf (COTS)…”
What’s with the stupid acronym?
Anyhow, good look to them. It’s exactly what a lot of people have been wanting.
what’s to port? it’s scripts that are customized to THEIR distro, which is different than OTHER distros.
write your own scripts to customize and integrate apps….this one isn’t even applicable to other distros due to the way Lycoris has been customized.
anybody that keeps talking about porting it doesn’t understand what it is.
lycoris doesn’t need the vendors to support it necessarily, as they can write their own scripts to do it where possible.
the author did have a point though, and I think this is something that all distros should try to do….but Lycoris’ AI2 would do no good for any other distro.
Is we need a universal package integration system that can be ported to all distrobutions to allow software designers the ability to create installation packages that work regardless of distribution. The user interface may vary greatly, the underlying system may vary greatly – all I’m talking about is a piece of software that takes a standardized installation package and translates it into a fully installed program within it’s own specific distrobution, fully managable and maintainable and useable by the user as if the application/program/software was precicely designed to be installed and executed on that specific distrobution itself.
You can make a distro as complicated or simple or as craftable or as secure or customizable as you want – but until you design a standard packaging and integration layer that tells an installation package ‘Hey! This is how we do things around here, this is where everything goes, this is where everything you want is.’ and installs it accordingly, Average Joe is always going to find it troublesome to try out new software for the heck of it because it’s such a hassle.
Mind this isn’t for power users who don’t have any trouble sorting out dependancies (let alone understanding them), compiling programs from source, configuring text files, and so on. I’m not even speaking for the distrobutions you might work on. I’m just saying for distrobutions where the target audience is Average Joe, like Lycoris or Xandros or Lindows – Or even Mandrake, Suse, Fedora for that matter possibly, that some sort of universal standard might be a good idea.
Just my humble thoughts.
Read the first paragraph under the “will it fly” section. The problem is Lycoris’ marketshare. If I was a COTS vendor – btw, that’s a standard industry acronymn – why would I bother with AI2 unless most of my customers were going to use Lycoris (which is unlikely)? If I just ignore AI2, then my app will work just as well on Lycoris as on other distros. I’ve not gained much if the time I spend on AI2 scripts is wasted because most of my customers are using Mandrakelinux, for example.
If Lycoris reimplement the AI2 API on Mandrakelinux, SUSE, RedHat and Debian, to name a few, then my customers can use any of these distros and I’m not going to have to force them to choose Lycoris Desktop/LX. Instead, I can offer them an app that has a big competitive advantage over those of my rivals with very little effort on my part. It just makes sense to use AI2 if it works on most major distros.
That’s hypocritcal – you personally said you would not share that with other distributions! I have logs from the IRC conference in #lycoris.
Anways… the way I see it if Lycoris has AI2, it’s a reason to use and develop the application on Desktop/LX. It makes the integration for other app companies much easier.
Each Linux vendor has something to give it an edge… this is Lycoris’
No, I said I wouldn’t share the source with other distros. That doesn’t preclude a port.
Lycoris’ original intention may well have been what you describe, but they would be stupid to throw away a good idea (porting and then dominating the COTS integration market) because they didn’t come up with it themselves. It all depends on how stubborn they want to be wrt maintaining their original vision for the technology.
That’s a very good point.
However, it’s only advantageous for Lycoris users if COTS vendors are happy to AI2-enable their apps.
The question then becomes: how does Lycoris build a viable developer community around AI2? The answer is by broadening the potential customer base available to vendors. There are three ways to achieve this:
1. Dominate the desktop OS market
2. Do all the work themselves for every COTS app out there
3. Extend AI2 so that it works with other distros
The second two options are the most viable, and of those two it is the third that makes the most sense over the long term.
Sorry.. but COTS is a dumb word.
But your questions and proposed solutions at applicable in any Vendor market. It applies to all and not just this software that Lycoris has developed.
I guess you will have to wait and see what happens. AI2 does have potential, I agree, but that is not a requirement for it to be handed out to all…
Also, since when have to Linux vendors ever agreed on a technology? GTK/QT D-BUS/DCOP list goes on and on…
Tell me about it! COTS is dumb and Linux vendors have a hard time agreeing.
Porting does not necessarily mean handing it out. The API backends can be closed source or otherwise restricted, possibly with an Aladdin-like license.
The beauty of porting AI2 is that the Linux distro developers don’t need to agree. Lycoris can maintain the backends and adjust them as the distros adjust their own standards. The same goes for the app vendors, they can maintain any old standards they want, and let AI2 provide the glue to make their non-standard apps integrate with the various distros their users run them on.
As for handing it out to all, Lycoris needs to act quickly because the technology involved (scripting) is not exactly new and AI2 can and will be implemented by a competitor along the lines I have argued for unless they seize the opportunity presented by being first to market with their integration infrastructure. If Novell or IBM or Sun decides they like this idea, and get something running along those lines, it will be Lycoris that loses.
how do you port scripts that are dependant on features that only lycoris has or chooses to use?
if other distros want to try similar things, I don’t think there’s anything stopping them…..
Two things: Firstly, what features are Lycoris-exclusive? My Linux System is pretty special, true, but its basic functionality is largely duplicated in the kio devices slave. And every distro has a control panel of some kind, and a start menu.
Secondly, if I am writing a script that is trying to put icons in a start menu, I need to know where the start menu is. I can run a detection subroutine to find it, or I can hard-code it, or I can detect the distro and work it accordingly. Whatever, it’s hard work.
Now, with AI2 I can call the API, and create a link to [startmenulocation]MyAppGroupMyApp.desktop, knowing that the API will convert [startmenulocation] to the actual start menu location on the target distro.
But if AI2 only works on Lycoris, what have I gained? I still have to detect all those other distros. If, on the other hand, AI2 is present in Debian, for example, I don’t need to code a special set of instructions for Debian — the AI2 backend will do all the work.
Sorry, two clarifications – when I say “what features are Lycoris-exclusive?” I mean, what features are Lycoris-exclusive in this context? The only thing the API covers according to the white paper are My Linux System, the control panel and the start menu.
Secondly, I’m not talking about porting the scripts. The scripts should work on all target platforms anyway, they are just shell scripts. I am talking about porting the API – the stuff that tells the scripts where the device access systen, the start menu and the control panel are – and the engine that runs the recognizer and enabler scripts automatically during the user login.
Regarding menus, I thought KDE and GNOME had agreed using the same configuration files for their menus so when some program adds entries to the menu system, it appears in both KDE and GNOME. Is this correct?
A large amount of this API is already deprecated thanks to freedesktop.org already implementing standards for the locations and formats of certain items like the standardized location for *.desktop menu entires. Why would I want to use AI2 when I can just create the *.desktop and not have any improved integration or anything?
I really think some of these stupid companies, like Lycoris, should get out and look at what is actually happening with the open source software they are using. Freedesktop is already working on many integration projects, such as hardware integration and detection with HAL and DBUS. We don’t need an API that is specific to Lycoris to do it.
The desktop Lycoris uses, KDE, has a framework called KConfigXT that stores desktop-wide settings. Do Lycoris know about this, or do they even know about Freedesktop? Probably not, because they don’t contribute or get involved with KDE and they just don’t damn well make themselves aware of what’s going on.
Although the method, shell scripting, is hardly new or innovative, the idea that integration difficulties caused by the proliferation of competing Linux platforms could be overcome with a few simple scripts is revolutionary.
That’s rubbish, as it is only going to work with Lycoris. This does not provide them with competitive advantage, and should be submitted to a group of people such as Freedesktop.
Although much of the integration that is possible with AI2 can be replicated with little difficultly on non-Lycoris platforms, some tasks, such as integration with the desktop start menu and control panel, are highly dependent on the underlying operating system.
KDE and Gnome have a common, shared Freedesktop format for desktop and menu entries, and have done for some time. Certainly, if the distributor hasn’t done anything stupid menu entries for KDE will certainly work for any KDE desktop, so I don’t get that. Get out there and find out!
“It can’t really be ported to other operating systems due to compatibility issues. It doesn’t need to be, and it’s not going to be.”
Coming from a company that wouldn’t have any business without free and open source software, I find that statement absolutely ludicrous. It could very easily be ported to other systems, but this is just scaremongering on his part because he thinks he’s doing a Microsoft and making something that only works on Lycoris. Considering the market that Lycoris has, this obviously isn’t going to fly. This tells me that Lycoris is dead as a company, and certainly has no long term future.
I’ve tried to download the whitepaper on it, and I’ve even got to register to do that. You wouldn’t think Lycoris wanted anyone’s business, would you?
I’ve always liked lycoris, since it was Redmond Linux, I really like what they’re trying to do: simplify the user experience of linux, so that the user can get on and do the things they want and leave all the complex stuff up to the OS to sort out (i.e. how every OS SHOULD be!)
AI2 sounds like a great idea, sounds like just the thing the Linux community needs to get the os more usable and more mainstream (let’s face it – it’s still a nerd’s/geek’s OS)
I currently use Suse 9.1 and it’s great, as long as I don’t want to install any new software (dependancy hell), compile from source (always missing headers or libraries) or heaven forbid try to uninstall something (more dependency hell!)
If this can take some out of the pain of installation and operation – i.e. default saving to my directory of choice (or specific to the OS) then fine.
would it make me switch my linux distro… sure would, anything that makes Linux easier would.
I wish them all the best of luck.
Read my lips: This Is For Closed Source Commercial Software. This is for stuff that Won’t Be Changed at source level to fit emerging standards.
Look at Win4Lin or VMWare and tell me again that they integrate well with the current crop of so-called standardized desktops. They don’t.
I’ve tried to download the whitepaper on it, and I’ve even got to register to do that. You wouldn’t think Lycoris wanted anyone’s business, would you?
So? Red Hat make you register for webcasts; Novell makes you register if you try to download SUSE Personal from their website. They are just trying to track people’s interest – what’s wrong with that?
Read my lips: This Is For Closed Source Commercial Software. This is for stuff that Won’t Be Changed at source level to fit emerging standards.
So what? Does AI2 itself take advantage of emerging standards, and will it in the future?
Software is being changed all the time by commercial software developers to fit emerging standards. To be truly integrative in the manner that people want, existing software like Win4Lin or VMWare (new versions all the time) will have to be improved to take advantage of emerging standards. Are you suggesting that these people aren’t going to improve their software? With a good support structure, it will be far easier to upgrade people and improve their systems than it currently is with Windows.
Since it is distributors that package software, rather than people installing it themselves, the current need for AI2 in this context is limited (and won’t solve anything, as the software needs to take advantage of the system to be really integrated). If you’re talking about downloading and installing files over the internet, like people do with Windows, then it is going to take a heck of a lot more than AI2. It will need the full support of all the free software, projects and companies that contribute that is used within a standard distribution. That’s a different ball-game. I wish people would get away from this stupid attitude that you can do development that affects free software completely in-house.
The legitimate question a customer will ask is “Will it work between distributions?”. In a Linux desktop world, in case you hadn’t noticed, there is more than one distribution. If Lycoris does it all better than anyone else, people will stick with it.
Look at Win4Lin or VMWare and tell me again that they integrate well with the current crop of so-called standardized desktops. They don’t.
Why don’t you ask Win4Lin and the VMWare people why their software isn’t integrated? What’s wrong, what could be improved? Integration is also a loaded word, so it really depends what you mean by it. Rather than going off and developing something in-house, Lycoris needs to start talking to people first.
That’s what a body of work like Freedesktop is for – so that we have common technology that will work between distributions. That is what is required for Linux desktops to really succeed. If we have distribution specific technology then desktop Linux fails, just as desktop Unix failed in the late 80s and early 90s. The Lycoris initiative isn’t going to help here because it is Lycoris specific.
So? Red Hat make you register for webcasts; Novell makes you register if you try to download SUSE Personal from their website. They are just trying to track people’s interest – what’s wrong with that?
Then get people to fill out a simple questionnaire, but don’t ask for e-mail addresses to send people spam with. The point is that if this is a whitepaper that is designed to garner interest, industry support and support from free software projects (which is critical) then let people read it. Registering for a download of a product or a webcast is different. A whitepaper on a technology should be open to allow people to read it, as Suse and Red Hat actually do.
You really don’t understand the free software approach that the software the Lycoris distribution is made out of is actually built on. If you have a Microsoft attitude here, then you will fail miserably. You need to understand and get a feel for what is required in a free software world. Release the technology to other distributions, but make sure you do it better than anyone else. Technology protectionism isn’t going to work, I’m afraid.
Commercial or proprietary software is definitely not going to disappear (I’m not one of those people), but you need to look at the core free software that is used in the right way to succeed. Believe it or not, I’m worried about the future of Lycoris.
So what? Does AI2 itself take advantage of emerging standards, and will it in the future?
There’s no need for it to do so – it’s trying to bind non-standard applications to a target distribution. Whether the target distribution complies with emerging standards is only a matter of interest.
Software is being changed all the time by commercial software developers to fit emerging standards. To be truly integrative in the manner that people want, existing software like Win4Lin or VMWare (new versions all the time) will have to be improved to take advantage of emerging standards. Are you suggesting that these people aren’t going to improve their software? With a good support structure, it will be far easier to upgrade people and improve their systems than it currently is with Windows.
Emerging standards are just that – emerging, and not fully fleshed out. Look at http://www.freedesktop.org/Standards/Home and you’ll see that most of what’s relevent to this discussion are draft documents, some with de facto acceptance but many without.
So while Open Source developers are likely to do as you suggest, and follow the standards, commercial developers will not. Why would I spend thousands of dollars making my application conform to a draft specification, when my competitor is spending his money making a cool new feature? In the end, after all, the integration features of an application, however useful, are really only the “on ramp” to the application itself.
The difference Linux has with Windows is that with the Windows Installer package, a developer can create a package that integrates tightly with all versions of Windows fairly easily. With InstallShield it’s a no-brainer for even mediocre developers.
With Linux, tight integration requires a lot of work, and developers have yet to see how they could get any return from their investments in this area. Users can see the need, but obviously software vendors have decided that their “good enough” strategy works well enough for them to ignore the need for better integration.
AI2 could change this paradigm, because developers could achieve the tight integration users want without having to hard-code changes for each target platform – some will no doubt be standards compliant, some will always not be.
I wish people would get away from this stupid attitude that you can do development that affects free software completely in-house.
How does AI2 affect free software in any meaningful way? All it does is bind commercial, non-free software to some Linux distributions (and atm only one), and mostly just by manipulating configuration files AFAIK.
Why don’t you ask Win4Lin and the VMWare people why their software isn’t integrated? What’s wrong, what could be improved? Integration is also a loaded word, so it really depends what you mean by it. Rather than going off and developing something in-house, Lycoris needs to start talking to people first.
I think Lycoris have hit the mark bang on with this effort. I’ve outlined why vendors are unlikely to spend time and effort improving on their integration at source level: basically it boils down to a cost/benefit paradigm. Unless you can change the costs, which AI2 attempts to, or the benefits, by vastly increasing Linux’s market share, this paradigm is unlikely to change any time soon.
That’s what a body of work like Freedesktop is for – so that we have common technology that will work between distributions. That is what is required for Linux desktops to really succeed. If we have distribution specific technology then desktop Linux fails, just as desktop Unix failed in the late 80s and early 90s. The Lycoris initiative isn’t going to help here because it is Lycoris specific.
That’s why I am arguing for a port, so that AI2 becomes more than the lock-in technology it currently is. It doesn’t really matter to commercial developers whether the whole system is open source or not, just that the API is documented well enough that they can do what they have to do – and, of course, that the system works.
That’s something to strive for. </sarcasm>
It does exist for linux as well by the way and is distro agnostic. Given how popular it is with commercial developpers I don’t have much trouble figuring what they’d choose between them and AI2.
AI2 and InstallShield are different. InstallShield, well installs the software and such. AI2 on the other hand, configures the software and integrates it into Desktop/LX (for example, adding the Crossover Office and Cedega “Harddrives” to My Linux System.
There’s no need for it to do so – it’s trying to bind non-standard applications to a target distribution. Whether the target distribution complies with emerging standards is only a matter of interest.
Then people are not going to be interested, nor are other distributions and free software projects going to use it. It’s a very, very different world to Windows, I’m afraid. Lycoris also do not have the userbase to make it meaningful, nor will they ever in an open source world.
So while Open Source developers are likely to do as you suggest, and follow the standards, commercial developers will not. Why would I spend thousands of dollars making my application conform to a draft specification, when my competitor is spending his money making a cool new feature?
Because in an open source world, the one that conforms to standards works the best in any given set of circumstances and in different distributions and is something you just can’t wrap your head around. You innovate within those standards, and if you break them, nothing works everywhere and no one uses your software. Red Hat have learned that the hard way when they created their Bluecurve desktop, as have nVidia and their drivers and as ATI are doing with theirs.
Commercial developers are not the ones who create the free software that distributions like Lycoris rely on. Unless commercial software plays well with the existing core free software, and developers are able to run it well, the free software is not going to play nice with commercial software. Result? Commercial software becomes unreliable, doesn’t work everywhere and people use something else.
The big distributions like Suse and Red Hat are investing in and contributing to free software – there’s a hint. If there is a free and open spec version of AI2, guess what will happen to Lycoris and AI2?
It’s a whole different ball game, and unless companies like Lycoris understand it, they’re not going to last very long.
AI2 could change this paradigm, because developers could achieve the tight integration users want without having to hard-code changes for each target platform
Each platform? This will only work on Lycoris.
I’ve outlined why vendors are unlikely to spend time and effort improving on their integration at source level: basically it boils down to a cost/benefit paradigm.
So why do developers put effort into storing registry settings or taking advantage of platform specific features on Windows? Because not to do so would lead to a level of non-integration that would unthinkable.
Then people are not going to be interested, nor are other distributions and free software projects going to use it. It’s a very, very different world to Windows, I’m afraid. Lycoris also do not have the userbase to make it meaningful, nor will they ever in an open source world.
It’s a very different world for geeks. For everybody else, the requirements of computing are the same as with Windows.
Because in an open source world, the one that conforms to standards works the best in any given set of circumstances and in different distributions and is something you just can’t wrap your head around. You innovate within those standards, and if you break them, nothing works everywhere and no one uses your software. Red Hat have learned that the hard way when they created their Bluecurve desktop, as have nVidia and their drivers and as ATI are doing with theirs.
Last time I looked Red Hat was the market leader, and there was a huge outcry when they stopped their consumer Linux line, BlueCurve or no BlueCurve. And NVIDIA and ATI? Don’t make me laugh! Their stuff is closed source and their installers are about as standards-compliant as a cow on the moon. Okay, slight exaggeration maybe but true.
Commercial developers are not the ones who create the free software that distributions like Lycoris rely on. Unless commercial software plays well with the existing core free software, and developers are able to run it well, the free software is not going to play nice with commercial software. Result? Commercial software becomes unreliable, doesn’t work everywhere and people use something else.
That’s true, but once again this is not aimed at Open Source software developers. I agree that commercial software should play nicely with others, but emerging standards are, as I have already outlined, unlikely to form the glue that binds them to their closed source counterparts. Accepted standards, or hacks like this one, are another matter, the former because they are just accepted, like the Windows registry, and the latter because they are easy and/or cheap.
The big distributions like Suse and Red Hat are investing in and contributing to free software – there’s a hint. If there is a free and open spec version of AI2, guess what will happen to Lycoris and AI2?
All their intellectual property will dissipate in an instant and they will be left with the grandeur of being the inventors of a clever API — and no money. AI2 will fly and Lycoris will disappear.
Each platform? This will only work on Lycoris.
If they port, it will be each platform. That was the “could” part of my original statement.
So why do developers put effort into storing registry settings or taking advantage of platform specific features on Windows? Because not to do so would lead to a level of non-integration that would unthinkable.
I’m glad you mentioned the registry. The main reason the registry is so ubiquitous in the software world is that it is a (fairly) static API and storage tool, with iron-clad compatibility across platforms. A program that interacts with Windows XP’s registry, for example, will almost certainly work with Windows 2000 and Windows NT 4.0; it will probably work with Windows 9x/ME as well, unless is uses some of the more esoteric keys.
That ubiquity doesn’t apply to Linux. Try writing a script that alters some KDE settings on say Debian and Red Hat. Chances are that you’ll need to recode the portions dealing with the location of the configuration files and maybe the actual keys as well. Even though standards are emerging, they are only emerging, and distributions have yet to implement them, if they ever will. So we are left with major problems.
Commercial application developers can do what you propose and sit and wait for Godot, Godot in this case being harmonised standards, or they can use something like AI2 that works in the here and now (at least if Lycoris ports it across distributions). Which to choose? Hard choice, I know, but AI2 sounds pretty good by comparison to waiting. Because I think that standardisation is a long time away, if indeed it ever happens. And if Lycoris ports, it’s quite possible that by the time the other distributions finally get around to agreeing on the various standards, AI2, or something else like it, will have become the de facto standard simply because it was on the market first and because it is being used by so many commercial developers.
It’s a very different world for geeks. For everybody else, the requirements of computing are the same as with Windows.
So it is, but the OS is not built on proprietary components. Therein lies the difference, and is why you don’t understand.
Last time I looked Red Hat was the market leader, and there was a huge outcry when they stopped their consumer Linux line, BlueCurve or no BlueCurve.
Has nothing to do with what I’ve written.
And NVIDIA and ATI? Don’t make me laugh! Their stuff is closed source and their installers are about as standards-compliant as a cow on the moon. Okay, slight exaggeration maybe but true.
So what? Oh wow – they’re closed source! They have learned that they can no longer do testing totally in-house as they are interfacing in an environment full of open source components. They need people who understand them. That’s why ATI still have some way to go. There are no standards for installers yet, and if they’re not standards compliant, neither is AI2.
I’m glad you mentioned the registry. The main reason the registry is so ubiquitous in the software world is that it is a (fairly) static API and storage tool, with iron-clad compatibility across platforms.
Not interested. My point was that you said it was too expensive for developers to integrate with a platform. It clearly isn’t if the standards are there – which they are, especially for menu and desktop installation entries.
That ubiquity doesn’t apply to Linux. Try writing a script that alters some KDE settings on say Debian and Red Hat. Chances are that you’ll need to recode the portions dealing with the location of the configuration files and maybe the actual keys as well.
No you don’t. The environment variables tell you where KDE is installed, the rest is automatic as it is simply KDE. I don’t know what on Earth you mean by ‘keys’.
Commercial application developers can do what you propose and sit and wait for Godot, Godot in this case being harmonised standards, or they can use something like AI2 that works in the here and now (at least if Lycoris ports it across distributions).
Or they can simply use existing standards now. Hard choice…