I've used Linux for approx. 4 months now. From what I see in my brief time, I would say Linux needs:
---A common packaging/distribution system for all distros- mostly for the applications. I don't know, but I hope any application written for Linux will run on any distro put out there. If not, it should be possible.
---A common GUI-based development IDE that can be used for all distros. All I found so far is GCC and Eclipse. Eclipse is not ready for prime time in the Linux world, from the reviews I read.
---A common front. The Linux world is very fragmented, as most of us know. When I read up about a distro or sometimes an application, the usual read is: Yes, this is the best Linux ever!! Then upon further reading, I find that it's a dog compared to newer, more stable distros. This does not really happen with Windoze at all. Granted: slimmer distros have their place, with older computers, and possibly in the future with specialized applications as mobile phones, etc (if not being done already). The Distro-wars have to stop. The bickering and snipping at each other is tiring to read. Are we not all on the same team? I know Redmond is (one integrated team).
What Linux Needs
There are no Distro Wars, but no we aren't on the same team - each programmer is a free agent. What you see is the free market in action.
Would McDonald's be better if Burger King and Wendy's closed? Maybe if we all decided to work at McDonald's, their food would magically become healthier, tastier, cheaper, and nobody would ever have the desire to eat anything else.
---
Programmers are not fungible units of production, nor is it necessarily true that adding more programmers to any given project makes it get done quicker or better. In fact the opposite is more likely. The only choice is between having one sucky distro, and a dozen sucky distros. More is better because at least they suck in different ways.
Would McDonald's be better if Burger King and Wendy's closed? Maybe if we all decided to work at McDonald's, their food would magically become healthier, tastier, cheaper, and nobody would ever have the desire to eat anything else.
---
Programmers are not fungible units of production, nor is it necessarily true that adding more programmers to any given project makes it get done quicker or better. In fact the opposite is more likely. The only choice is between having one sucky distro, and a dozen sucky distros. More is better because at least they suck in different ways.
-
danperecky
- Posts: 3
- Joined: Thu 12 Aug 2010, 19:04
I think the term is coined: 'divide and conquer'.
My post was not to say that different distros are necessarily bad, but that all Linux Applications should run on any distro (not sure- may be happening now -at least in theory), Packaging should not be so confusing - Yast, rpm, tar... - a common App packaging tool would be great, and a common developer environment with an IDE - for all Linux distros in use. This would give the Linux community incredible cohesion, and promote Linux to many new levels.
My post was not to say that different distros are necessarily bad, but that all Linux Applications should run on any distro (not sure- may be happening now -at least in theory), Packaging should not be so confusing - Yast, rpm, tar... - a common App packaging tool would be great, and a common developer environment with an IDE - for all Linux distros in use. This would give the Linux community incredible cohesion, and promote Linux to many new levels.
Divide and conquer only works if your enemy was a united force in the first place. A better analogy is the "War on Drugs" or the "War on Teen Sex", where your enemy is actually human nature itself.
If you mean we need a Visual Basic for Linux, I won't disagree.
---
Packaging, the formats are already converging - but it won't help as even minor differences in how the OS is setup and what versions of libraries you have break things. Your best bet is to use interpreted languages, or build statically - but you pay with size and performance.
It's like wishing that you could put the same parts in every make of vehicle: sure if every car had the exact same engine - good luck with that.
If you mean we need a Visual Basic for Linux, I won't disagree.
---
Packaging, the formats are already converging - but it won't help as even minor differences in how the OS is setup and what versions of libraries you have break things. Your best bet is to use interpreted languages, or build statically - but you pay with size and performance.
It's like wishing that you could put the same parts in every make of vehicle: sure if every car had the exact same engine - good luck with that.
That is maybe a little simplistic jemimah. What we would really like is for all the different engines to fit the same mount points and use he same wiring. Now THAT would be choice.............jemimah wrote: It's like wishing that you could put the same parts in every make of vehicle: sure if every car had the exact same engine - good luck with that.
Utopian?
Spup Frugal HD and USB
Root forever!
Root forever!
-
Bruce B
One thing I think Linux needs is competent dedicated users.
Computing is about performing tasks.
With Windows I'd learned how to perform about any task I wanted
to.
Basic Linux was pretty easy for me because of earlier experience
with other operating systems.
However, I didn't know how to perform all the tasks which I
knew how to perform in Windows.
The temptation was to boot into Windows and perform the task.
Maybe after booting into Windows, I'd stay in Windows.
The problem is I wasn't really learning Linux like I had intended.
So, I deleted Windows, thus forcing a situation whereby if I wanted
to perform a task, I had to do it in Linux.
Eventually, I learned how to do all my tasking in Linux.
Funny, but after really learning Linux well, I don't think Linux 'needs'
much of anything it doesn't already have or isn't available.
Computing is about performing tasks.
With Windows I'd learned how to perform about any task I wanted
to.
Basic Linux was pretty easy for me because of earlier experience
with other operating systems.
However, I didn't know how to perform all the tasks which I
knew how to perform in Windows.
The temptation was to boot into Windows and perform the task.
Maybe after booting into Windows, I'd stay in Windows.
The problem is I wasn't really learning Linux like I had intended.
So, I deleted Windows, thus forcing a situation whereby if I wanted
to perform a task, I had to do it in Linux.
Eventually, I learned how to do all my tasking in Linux.
Funny, but after really learning Linux well, I don't think Linux 'needs'
much of anything it doesn't already have or isn't available.
Sure, but how exactly are you going to build a Smart Car if the engine from the F350 had to fit inside? By setting constraints, you limit what is possible.DaveS wrote:That is maybe a little simplistic jemimah. What we would really like is for all the different engines to fit the same mount points and use he same wiring. Now THAT would be choice.............jemimah wrote: It's like wishing that you could put the same parts in every make of vehicle: sure if every car had the exact same engine - good luck with that.
Utopian?
What makes Linux unique is the extreme flexibility that results from the fact that it's entirely open source. Not only is it impossible to prevent people from inventing and distributing new and incompatible Linuxes, such a concept is antithetical to the anti-authoritarian philosophy that allowed Linux to be created in the first place.
-
Pizzasgood
- Posts: 6183
- Joined: Wed 04 May 2005, 20:28
- Location: Knoxville, TN, USA
Binary packages are non-portable. That is how they are. Accept it.
If you want portability, you stick with source packages. Half the point in open-source is portability. (The other half being hackability, which includes all the freedom business.)
No, it isn't necessarily pretty. Yes you have to get your hands a little bit dirty.
Car analogies? Ok. Ever changed a tire? Those are fairly compatible standardized parts. You still have to make sure you get the right size for your car, and you have to do a bit of manual labor to swap it in. Might get a little dirty. But overall it isn't a very difficult process, other than when you get a stuck lug nut or have issues dealing with the hub cap. Most programs are much less hassle to compile than changing a tire is.
People are not asking for standard parts - which we already more or less have. They are asking for automatic parts. They don't want to change their tire. They want to push a button on their fancy remote-control key-fob deal and have the tire change itself.
We don't have that yet. The main element missing is a way to deal with dependencies when compiling. That still has to be done the hard way. Sometimes it's very annoying. What we need is a good way to find out which dependencies we need, without having to run ./configure and find out from the error messages. Because usually it stops at the first missing dependency. So you wander off to install that (finding out that it requires some dependencies of its own). When you finally get it done, you get a little farther only to find out you need another dependency.
Of course, you can often find them listed in the documentation. But not always. It would be nice if you could run "./configure --deps" and have it print out a list of what is needed, the version ranges, whether it's mandatory or optional, and maybe even a URL to the web page of the package.
I'm not talking about the entire dependency tree, just the first order stuff that the package needs. You'd still find out about second-order dependencies when looking at the first order dependencies. But seeing all of the first-order ones at once, in a standardized scriptable way would be great. It would make it much simpler to automate the entire process.
A lot of that information is already stored somewhere, otherwise ./configure wouldn't be able to yell at you about missing dependencies. Just need a better way to access it. Detect the calls to pkg-config or something.
If you want portability, you stick with source packages. Half the point in open-source is portability. (The other half being hackability, which includes all the freedom business.)
No, it isn't necessarily pretty. Yes you have to get your hands a little bit dirty.
Car analogies? Ok. Ever changed a tire? Those are fairly compatible standardized parts. You still have to make sure you get the right size for your car, and you have to do a bit of manual labor to swap it in. Might get a little dirty. But overall it isn't a very difficult process, other than when you get a stuck lug nut or have issues dealing with the hub cap. Most programs are much less hassle to compile than changing a tire is.
People are not asking for standard parts - which we already more or less have. They are asking for automatic parts. They don't want to change their tire. They want to push a button on their fancy remote-control key-fob deal and have the tire change itself.
We don't have that yet. The main element missing is a way to deal with dependencies when compiling. That still has to be done the hard way. Sometimes it's very annoying. What we need is a good way to find out which dependencies we need, without having to run ./configure and find out from the error messages. Because usually it stops at the first missing dependency. So you wander off to install that (finding out that it requires some dependencies of its own). When you finally get it done, you get a little farther only to find out you need another dependency.
Of course, you can often find them listed in the documentation. But not always. It would be nice if you could run "./configure --deps" and have it print out a list of what is needed, the version ranges, whether it's mandatory or optional, and maybe even a URL to the web page of the package.
I'm not talking about the entire dependency tree, just the first order stuff that the package needs. You'd still find out about second-order dependencies when looking at the first order dependencies. But seeing all of the first-order ones at once, in a standardized scriptable way would be great. It would make it much simpler to automate the entire process.
A lot of that information is already stored somewhere, otherwise ./configure wouldn't be able to yell at you about missing dependencies. Just need a better way to access it. Detect the calls to pkg-config or something.
[size=75]Between depriving a man of one hour from his life and depriving him of his life there exists only a difference of degree. --Muad'Dib[/size]
[img]http://www.browserloadofcoolness.com/sig.png[/img]
[img]http://www.browserloadofcoolness.com/sig.png[/img]
-
sickgut
- Posts: 1156
- Joined: Tue 23 Mar 2010, 19:11
- Location: Tasmania, Australia in the mountains.
- Contact:
i couldnt agree more, the way things are going windows is getting more dumbed down, soon there wont be a distinction between the interface on a mobile phone and the main computer. Also the whole thing is designed so that users will spend more money by visiting other money making sites or using their programs. Computers for teenagers are just becoming facebook/ youtube machines.jemimah wrote:By trying to set Linux up as the competition to Mac and Windows, you end up removing most of the advantages of Linux.
Linux is great because of the diversity and distributed development. The software is often developed by the actual end users - not by committee, and not for the purpose of making a profit. The result is power and flexibility.
However that's the very reason it kind of sucks as a desktop OS. There's no overall vision. No cohesiveness in application design. No user acceptance testing.
It's kind like a B movie. The awesomeness comes from the fact that people like you made it, on a budget like 1/1000th of the size of a blockbuster. So while the effects kind of suck, and the actors are inexperienced and awkward, the plot is more imaginative and satisfying because it hasn't been sanitized by marketing, dumbed down to appeal to the lowest common denominator, or censored to avoid offending the sensitive.
Want to see what commercial desktop Linux looks like? Check out MeeGo, ChromeOS, and Android (and Ubuntu to some extent). They've made it more user friendly by limiting what you can do with it. It's not intended to be customized much. I imagine functionality will increase over time, but I'll still be running Puppy - quirks and all because I can make it mine and it appeals to my imagination.
i would hate to see linux dumbed down to that extent, android came from linux stock and is already dumbed down to that extent and its a pitiful sight to see. Maybe someone intelligent enough to know they want to do something other than surf the web would be intelligent enough to run a full OS. Or maybe easy mode and expert mode settings that change the gui and locking mechanics is the answer.
linux made by technical people for technical people can do technical things very well. However to do technical things you generally need a technical interface with alot of options. When you dumb linux down like android has you loose the ability to do all the technical things you wanted to do with linux to begin with.
i hope linux retains its technicality. We shouldnt make linux like windows, there will always be a learning curve to change from windows to linux. I would prefer to believe that people with an average IQ of 100 can actually learn to use a different interface.
by making linux more like windows we arent gaining anything. Maybe the result could be a free OS, but it would be free and crap. I prefer to think that free and good is a better option.
in the past linux has used windows as inspiration and adopted alot of things from it in a round about way, like the start menu in win 95 etc, compare that to most old style window managers. However i think linux has the advantage now and gained the higher ground. If windows died today, linux would continue to get better and better at the pace its already at. Thank you windows for a looksee to see what could be done, but we simply dont need you anymore.
This isn't exactly the free market which is about selling a commodity for the maximum profit. Many programmers on source projects are volunteers who give there time to add to the creative commons. Therefore they are not selling their work or even their time. Ones who get paid are paid not because of the programs but from other things like support and advertising. Therefore the programs aren't on the market at all. IMO open source works because you can not pay for creativity ie. paying someone more money will not make them more creative.jemimah wrote:There are no Distro Wars, but no we aren't on the same team - each programmer is a free agent. What you see is the free market in action.
what linux needs is for people to understand what it is for
those who have the spare time to learn how their computer works, and the patience to gradually solve complex problems will love linux.
those who just want to get on with their life and do what everyone else is doing with their computers will hate linux.
those who want to do both, will. because its not hard to lead a double life. puppy makes it easy to live a double life. just keep the cd in your kit
those who have the spare time to learn how their computer works, and the patience to gradually solve complex problems will love linux.
those who just want to get on with their life and do what everyone else is doing with their computers will hate linux.
those who want to do both, will. because its not hard to lead a double life. puppy makes it easy to live a double life. just keep the cd in your kit