PDA

View Full Version : Standards do not destroy freedom!



Dekkon
March 20th, 2009, 01:12 AM
There's quite a lot of people going around saying, if there are standards, that simply doesn't follow the meaning of open-source and freedom.

Standards, are just say, a default option, something that comes standard on a computer. Just because your distro followed a standard and had such backend installed, doesn't stop your freedom of changing.

You can still have your freedom, while us non-technical users, who use to computers for work, have are standards without looking up tutorials on how to install one freaking software.

The post is pointless, but I'm sorry, choice is good, and choice is bad, let's implement choice in a smart way, not redundant, quantity over quality, way.

cardinals_fan
March 20th, 2009, 01:24 AM
There's obviously something you're trying to say. Let it out. What do you mean by standards?

days_of_ruin
March 20th, 2009, 01:30 AM
Sorry I couldn't resist option number 3.

Dekkon
March 20th, 2009, 01:37 AM
There's obviously something you're trying to say. Let it out. What do you mean by standards?

Most people around here already know my views on the different choices of package managers, and the strain it puts on new users and the overwhelming feeling they must encounter, let's just leave it at that. =p


Sorry I couldn't resist option number 3.

I did have a third option but I forgot what it was gonna be, so I put that instead. I know there would them few people who couldn't resist, I sure wouldn't have.

Helios1276
March 20th, 2009, 01:40 AM
Surely without some standards we have only chaos! bedlam no less!?:popcorn:

Tibuda
March 20th, 2009, 01:42 AM
You mean standards as ODF or as OOXML?

Dr Small
March 20th, 2009, 01:44 AM
I live by standards, and so do my projects and code, even though they are not written. Standards are good, but it certainly does not stop innovation, nor restricts a users freedom, methinks.

issih
March 20th, 2009, 01:46 AM
My opinion is that the diversity of the linux ecosystem is a good thing, but that the job of a distribution is to produce a consistent coherent set of packages and make them work in a sensible and well integrated way. Ideally they should also stick to some actual standard definition of user interface guidelines for as many packages as possoble.

That way each distribution can be different, but also each one is internally consitent.

It isn't an overwhelmingly popular view though.

Dekkon
March 20th, 2009, 01:47 AM
You mean standards as ODF or as OOXML?

I'm talking about any standard, let's use ODF as an example though.

Most people complain about freedom, saying choice is a good thing, developers hate standards but non of them would complain about ODF, if there was something wrong with the format, developers would work to improve that standard, instead of creating twenty more formats with the same purpose.

I feel as if, the open-source community doesn't work close enough to create two great projects to tailor to different users needs, instead they create twenty half-*** projects and one or two "Doable" projects.

Open-source should be about collaboration, people working togethor, instead of the other option.

Uninformatively, nothing will be done to improve on this, things will remain the same, because developers will not even think about the options the users want, or come up with, they will only do what they think is best.

cardinals_fan
March 20th, 2009, 01:57 AM
Most people around here already know my views on the different choices of package managers, and the strain it puts on new users and the overwhelming feeling they must encounter, let's just leave it at that. =p

The fundamental problem is that a packaging standard cannot be created without quashing freedom and choice. How would you enforce the standard? How would you choose the system used?

Each packaging system offers something unique. I have yet to discover anything that works as well for me as Tazwok/Tazpkg on SliTaz. I feel that Tazwok has greatly increased my productivity by allowing me to perform the tweaks I need without too much wasted time or effort. Everything has its own uses.

I would not be opposed to a general recommendation to all devs to package their software in xyz format. As I see it, it's the job of a distro to package the apps in its repos, not the job of developers. Provided that an app is open source, packaging is up to distributions. I personally feel that closed-source apps should be delivered in non-distro-specific binaries. Scripts can be created by the distributions that automatically download each individual app (like the get-* scripts on SliTaz) and then packaged and added to the repos.

Dekkon
March 20th, 2009, 02:01 AM
The fundamental problem is that a packaging standard cannot be created without quashing freedom and choice. How would you enforce the standard? How would you choose the system used?

Each packaging system offers something unique. I have yet to discover anything that works as well for me as Tazwok/Tazpkg on SliTaz. I feel that Tazwok has greatly increased my productivity by allowing me to perform the tweaks I need without too much wasted time or effort. Everything has its own uses.

I would not be opposed to a general recommendation to all devs to package their software in xyz format. As I see it, it's the job of a distro to package the apps in its repos, not the job of developers. Provided that an app is open source, packaging is up to distributions. I personally feel that closed-source apps should be delivered in non-distro-specific binaries. Scripts can be created by the distributions that automatically download each individual app (like the get-* scripts on SliTaz) and then packaged and added to the repos.

As you see, this is a problem, which is why I didn't go into package management much. Instead of having a meeting with package management developers and brainstorming ideas, we went ahead and created 10 package management systems.

Think about the size of software if the market share jumped to 20%, it is just very redundant and time consuming to package, and maintain software 10* the amount we have now, something obviously has to be done, but it won't.

cardinals_fan
March 20th, 2009, 02:11 AM
As you see, this is a problem, which is why I didn't go into package management much. Instead of having a meeting with package management developers and brainstorming ideas, we went ahead and created 10 package management systems.

Think about the size of software if the market share jumped to 20%, it is just very redundant and time consuming to package, and maintain software 10* the amount we have now, something obviously has to be done, but it won't.
http://images.despair.com/products/demotivators/meetings.jpg

Sorry, I couldn't resist :). And there is some truth to that. The problem is that it would be very difficult to go up to a bunch of Debian devs and say "you know, I think you should really redo your whole packaging system to integrate this functionality". The 20,000 packages in Debian's repos represent a tremendous amount of work and are a significant asset. They won't throw away all of that.

Anyway, I think the proposal in my last post handles packaging pretty well. There is nothing wrong with having distros buck the trend and create their own system. It's important to realize that doing so requires them to make their own packages. Open source apps aren't difficult to package (if they are, the packaging system is defective). If I create a distro with a new packaging system, I accept the responsibility of packaging my own apps. And closed source apps can be distributed through universal binaries.

This solution applies to other stuff as well. Take file formats. OpenOffice is the recognized open source leader, and the vast majority of open-formatted files are in ODF. If I create a new word processor, I accept the responsibility of allowing the use of ODFs if I want to garner any supporters.

interllect
March 20th, 2009, 02:12 AM
Trick question?

Dekkon
March 20th, 2009, 02:16 AM
http://images.despair.com/products/demotivators/meetings.jpg

Sorry, I couldn't resist :). And there is some truth to that. The problem is that it would be very difficult to go up to a bunch of Debian devs and say "you know, I think you should really redo your whole packaging system to integrate this functionality". The 20,000 packages in Debian's repos represent a tremendous amount of work and are a significant asset. They won't throw away all of that.

Anyway, I think the proposal in my last post handles packaging pretty well. There is nothing wrong with having distros buck the trend and create their own system. It's important to realize that doing so requires them to make their own packages. Open source apps aren't difficult to package (if they are, the packaging system is defective). If I create a distro with a new packaging system, I accept the responsibility of packaging my own apps. And closed source apps can be distributed through universal binaries.

This solution applies to other stuff as well. Take file formats. OpenOffice is the recognized open source leader, and the vast majority of open-formatted files are in ODF. If I create a new word processor, I accept the responsibility of allowing the use of ODFs if I want to garner any supporters.

Image all the labor that went into packaging all that software, and even then, that software lags 2 years behind. I'm sure some of the devs that have half a brain could see there is easily a easier way to go on, then doing redundant work.

Also, what is the universal binary for Linux? Why don't we just use that?

cardinals_fan
March 20th, 2009, 02:26 AM
Image all the labor that went into packaging all that software, and even then, that software lags 2 years behind. I'm sure some of the devs that have half a brain could see there is easily a easier way to go on, then doing redundant work.
Packaging software shouldn't take a lot of labor. Could you clarify the statement "and even then, that software lags 2 years behind"?


Also, what is the universal binary for Linux? Why don't we just use that?
http://en.wikipedia.org/wiki/Executable_and_Linkable_Format

Every executable file fits this description. Proprietary software by definition only releases these binaries, not the actual source code. Most proprietary software on Linux is distributed with an installation script, which I approve of. A distro can add a package which downloads and runs the installation files and that script.

Giant Speck
March 20th, 2009, 02:31 AM
I voted number three because it's true.

:popcorn:

Dekkon
March 20th, 2009, 02:36 AM
Packaging software shouldn't take a lot of labor. Could you clarify the statement "and even then, that software lags 2 years behind"?


That statement wasn't realavent, I was just mentioning that Debian's packages are always further behind, but thats done on purpose for stability.



http://en.wikipedia.org/wiki/Executable_and_Linkable_Format

Every executable file fits this description. Proprietary software by definition only releases these binaries, not the actual source code. Most proprietary software on Linux is distributed with an installation script, which I approve of. A distro can add a package which downloads and runs the installation files and that script.

ITS A STANDARD, why certainly the developers would have complained about it and made duplicates by now, or is it just a good standard. Hmm, follows my point, I think so.

cardinals_fan
March 20th, 2009, 02:46 AM
ITS A STANDARD, why certainly the developers would have complained about it and made duplicates by now, or is it just a good standard. Hmm, follows my point, I think so.
Read the article. A system-wide executable format isn't an easy thing to change or screw around with. Emulation or a full library is required to run one format on another type of system. The "standard" isn't contractually binding, but it's almost impossible to ignore given the application of the technology.

I think this discussion would be more constructive if you could present a model or idea for standards - on any topic you wish. Packaging, document formats, manual pages... I'm game for any of it.

Dekkon
March 20th, 2009, 02:51 AM
Read the article. A system-wide executable format isn't an easy thing to change or screw around with. Emulation or a full library is required to run one format on another type of system. The "standard" isn't contractually binding, but it's almost impossible to ignore given the application of the technology.

I think this discussion would be more constructive if you could present a model or idea for standards - on any topic you wish. Packaging, document formats, manual pages... I'm game for any of it.

I was just giving ****. haha. :)

While I could certainly think of ways to do this, but I can't get real technical on the subject as I am not that experienced in that kind of development, or really any at all, just simple programs in PyQt4.

cardinals_fan
March 20th, 2009, 02:54 AM
While I could certainly think of ways to do this, but I can't get real technical on the subject as I am not that experienced in that kind of development, or really any at all, just simple programs in PyQt4.
It's totally okay to lack experience in development (believe me, the extent of mine is basic Perl scripts). But I would be careful asking for changes when you aren't in the shoes of those from whom you are demanding the changes.

Anyway, non-technical methods are welcome. My homework is done and most of my friends are offline right now, so I'd like something to do :)