Linux Standards Base released

[texon]

The news today (zdnet and elsewhere) mentions that a bunch of the other linux companies announced that they're working towards some kind of standard. Don't know much more than that. I assume that gentoo would be able to mesh with any such standard, just because its flexible enough to configure pretty much anyway one wants.

I didn't see gentoo specifically mentioned in the press releases, but is this standards effort something in which the gentoo community is going to participate?

texon

[Voltago]

is this standards effort something in which the gentoo community is going to participate?

Since LSB wants us to use RPM-based package management, i have my doubts about that.

[Carlo]

Since LSB wants us to use RPM-based package management, i have my doubts about that.

That's not correct.

Applications shall either be packaged in the RPM packaging format as defined in this specification, or supply an installer which is LSB conforming (for example, calls LSB commands and utilities).

http://refspecs.freestandards.org/LSB_2 ... stall.html

A lot of (even closed source) Software runs fine on Gentoo, so the compatibility can't be that bad, but I don't think we have the ressources to adhere to such specifications completely. Taking Gentoo's nature as a moving target into account, I would consider the attempt to do so more as a drawback, since it would consume a lot of time, which would be lacking for other tasks.

[deternal]

AFAIK LSB is a standardization of filesystempaths and a base set of packages that you would need - if that is in place you would have an LSB compliant system.

Now considering how gentoo systems are basically customized it seems logical that not all gentoo boxes can (nor should) be LSB compliant, however adhering to the systempaths and adding the possibility to fetch all the LSB basepackages at one time (ie emerge lsb, or a make-conf option such as x86LSB) would definitely be a nice feature.

Since I'm not a developer I do not know how much work the above mentioned would entail, but as a user I'd definitely appreciate the option. And most server admins probably will too (ie. commercial software companies can't just write off your problems on the distro - atleast not that easily, if you can claim atleast basic LSB compliance).

[kallamej]

Moved from Portage & Programming as it is not a support request.

[playfool]

The big news this time is standardising on C++, an area where the GCC guys protested the decision quite a bit as I understand.

I like LSB, we need standards to ensure that the user and the software maker gets the most out of Linux without having to support millions of path combinations and versions of standard software.

[skyfolly]

They said they did not want to repeat Unix's history(many forkings). So they had to standardise Linux.

And it is still underdevelopment, you can submit your own piece of work.

Create and maintain a list of criteria for evaluating and categorizing potential candidates.

[Evangelion]

is this standards effort something in which the gentoo community is going to participate?

Since LSB wants us to use RPM-based package management, i have my doubts about that.

No they don't. All they want is that rpm-packages will be installable in any Linux-system. And you can install RPM-system on Gentoo as well, so as far as packages are concerned, Gentoo can be compatible.

[playfool]

is this standards effort something in which the gentoo community is going to participate?

Since LSB wants us to use RPM-based package management, i have my doubts about that.

No they don't. All they want is that rpm-packages will be installable in any Linux-system. And you can install RPM-system on Gentoo as well, so as far as packages are concerned, Gentoo can be compatible.

As I can see we would fail on the init system, but aside that we are largely fine.. maybe some FHS stuff to consider though.

[aja]

is this standards effort something in which the gentoo community is going to participate?

Since LSB wants us to use RPM-based package management, i have my doubts about that.

No they don't. All they want is that rpm-packages will be installable in any Linux-system. And you can install RPM-system on Gentoo as well, so as far as packages are concerned, Gentoo can be compatible.

While that may be true now, note the following footnote in the section on Software Installation:

Supplying an RPM format package is encouraged because it makes systems easier to manage. A future version of the LSB may require RPM, or specify a way for an installer to update a package database.

Although Gentoo may be (and probably should be) incidentally compatible with many parts of the LSB (we may want to look at Filesystem, base packages, etc.), for as long as language like the above is in the LSB spec, I can't see it as being a priority (or anything but a nice-to-have) for Gentoo, or Debian for that matter.

And the cries from the GCC folks about "Why that particular C++ ABI" by no means should be ignored - they know what they are talking about, and enforcing a particular G++ ABI right now is a big problem.

Also, C has always been the lingua franca for system programming in Unix-like OSes, and I for one would need to see a pretty compelling argument to change that. An LSB requirement to include full RPM package support, a full suite of system libraries in C++, etc sounds like the words "streamlined", "low-footprint", "resource-efficient", etc. never made it into the standards conversation.

Sounds to me like yet another Unix specification that's trying to document all the major players (i.e. = "that's the way most of us are doing it" is much more important than "that's the best way to do it")

Yawn. I'll throw the new binder on the pile that has all the POSIX specs, the SUV spec, the various SUN extensions, blah blah

EDIT: Grammar correction, clarification of GCC stuff

[Archangel1]

Sounds to me like yet another Unix specification that's trying to document all the major players (i.e. = "that's the way most of us are doing it" is much more important than "that's the best way to do it")

Agreed. I don't like the emphasis on RPM - it's an open standard, but it's generally accepted that it doesn't work that well compared to, say, Portage. For a distro like Gentoo that does it better already, what exactly is to be gained by conforming to a standard like that?

[millenium_psyrax]

Also, C has always been the lingua franca for system programming in Unix-like OSes, and I for one would need to see a pretty compelling argument to change that.

Read any software engineering book published since who knows when and you'll have your argument (and I don't mean those "unbiased" software engineering in C books). The irony of your argument on this very point.

Sounds like how the major players do it, rather than the best way to do it

Do you honestly believe that C is the best language in which to develop 100,00+ line programs? Memory leaks and hanging pointers abound. C is better than Pascal and nearly all other languages from it's time and paradigm, but it sure does have its limitations. If you program in C and never have hanging pointers and memory leaks and you actually use dynamic memory allocation, then you are a master programmer, in fact, scratch that, grand master.

Have you ever built a distro from scratch, so many applications chuck their files in non-standard locations, you have to move those libraries and executables from one folder to another, then create symlinks to support programs that expect the library to be in the non-standard location. Patch this, patch that.

I think standards are a fantastic idea, how do you think the Unix companies lost to Microsoft, couldn't be bothered making standards or adhering to standards. Microsoft doesn't need standards, it owns the market (which is a real shame). I do think that if RPM becomes mandatory though, that is injust and politically motivated. But, Gentoo folk don't need to worry so much about that, because I would be quite surprised if Debian would ever comply, if it complies though, bang, RPM will be mandatory, no question. In fact, Gentoo really has no say in whether it is compliant or not, if the source programs all expect lib_gcc.so to be in a particular location, only a fool would move it to another location to be kinda cool.

[Ruzbeh]

What's so terribly wrong about RPM?

[Evangelion]

Sounds to me like yet another Unix specification that's trying to document all the major players (i.e. = "that's the way most of us are doing it" is much more important than "that's the best way to do it")

Agreed. I don't like the emphasis on RPM - it's an open standard, but it's generally accepted that it doesn't work that well compared to, say, Portage. For a distro like Gentoo that does it better already, what exactly is to be gained by conforming to a standard like that?

You are comparing apples to oranges. Portage is a whole system which fetches packaes, calculates dependancies and installs them. RPM is simply a method of packaging apps. More valid comparison to Portage would be something like up2date.

If you want to compare RPM's to something in Gentoo-world, more valid comparison would be ebuilds vs. RPM's.

[Ravenium]

I may have missed something by just skimming through the page, but I thought the minimal level of compatibility with this standard was "a system for tracking, versioning, installing, managing (etc etc)". Packages. Now, granted we don't directly support RPM in the way that redhat does it, but doesn't Portage comply with the IDEA of package management mentioned in the LSB spec?

I took away from the articles more that they're trying to make it so that you don't have to code x number of variations on your program or x number of installs for x number of linux distros. (The amusing thing that they focus on RPM is that right now you actually have to fuddle with your RPM if you're on an earlier redhat, fedora, or enterprise...or at least I did. Maybe it's not the best thing to use redhat's ideas as an example for a standard...)

And yes, the usual file paths for stuff, ie you'll always find things where the spec says you will. Who knows, I guess we'll have to see as it develops.

-Barry

[virtual]

Hi,

Well I belive standards are a good thing generaly speaking, BUT if you look at another big OS like M$ you can easily see the security threat when everyone does it the same way. 1 virus can spread to millons of pc's without much difficulty. Also I belive that standards cripple progress and inovation because you always have to follow the standard. Look at the user interface of W$-95 and all the way to someting-XP not much has changed. Now this is also a good thing but I do not want to see Linux suffer from it's own standard. Standards are to be used wisely and broken when needed.

[placeholder]

Well I belive standards are a good thing generaly speaking, BUT if you look at another big OS like M$ you can easily see the security threat when everyone does it the same way. 1 virus can spread to millons of pc's without much difficulty. Also I belive that standards cripple progress and inovation because you always have to follow the standard. Look at the user interface of W$-95 and all the way to someting-XP not much has changed. Now this is also a good thing but I do not want to see Linux suffer from it's own standard. Standards are to be used wisely and broken when needed.

Yes, but in the Windows world file permissions also do not exist, whereas in *NIX they do indeed exist which makes a virus's job impossible unless the person is dumb enough to run it as root.

Even if RPMs would "take over", you can use rpm2tgz in Gentoo in order to turn them into tgzs and then extract them. However, portage isn't in danger of going anywhere because it compiles stuff from the source, and seeing that having the source available is key in making "open source" software, the source shall stay.

[msimplay]

Sounds to me like yet another Unix specification that's trying to document all the major players (i.e. = "that's the way most of us are doing it" is much more important than "that's the best way to do it")

Agreed. I don't like the emphasis on RPM - it's an open standard, but it's generally accepted that it doesn't work that well compared to, say, Portage. For a distro like Gentoo that does it better already, what exactly is to be gained by conforming to a standard like that?

You are comparing apples to oranges. Portage is a whole system which fetches packaes, calculates dependancies and installs them. RPM is simply a method of packaging apps. More valid comparison to Portage would be something like up2date.

If you want to compare RPM's to something in Gentoo-world, more valid comparison would be ebuilds vs. RPM's.

not true since you can install binary packages on gentoo
and i have read its being further developed aswell

although we know the benefits of source compiling etc
sometimes it would be good to just install i686 binaries for things like KDE
which takes forever to compile

[KingPunk]

Well I belive standards are a good thing generaly speaking, BUT if you look at another big OS like M$ you can easily see the security threat when everyone does it the same way. 1 virus can spread to millons of pc's without much difficulty. Also I belive that standards cripple progress and inovation because you always have to follow the standard. Look at the user interface of W$-95 and all the way to someting-XP not much has changed. Now this is also a good thing but I do not want to see Linux suffer from it's own standard. Standards are to be used wisely and broken when needed.

Yes, but in the Windows world file permissions also do not exist, whereas in *NIX they do indeed exist which makes a virus's job impossible unless the person is dumb enough to run it as root.

Even if RPMs would "take over", you can use rpm2tgz in Gentoo in order to turn them into tgzs and then extract them. However, portage isn't in danger of going anywhere because it compiles stuff from the source, and seeing that having the source available is key in making "open source" software, the source shall stay.

..incorrect. they very much so do indeed exist. they're just less-known
because you don't have to screw with them half as much.
its a lesser known whichcraft. thats all

go into windows, right click any file. to properties, and then you'll see
the series of check boxes and whatnot. therein, lies your "file permissions"

:p

[episode96]

Well I belive standards are a good thing generaly speaking, BUT if you look at another big OS like M$ you can easily see the security threat when everyone does it the same way. 1 virus can spread to millons of pc's without much difficulty. Also I belive that standards cripple progress and inovation because you always have to follow the standard. Look at the user interface of W$-95 and all the way to someting-XP not much has changed. Now this is also a good thing but I do not want to see Linux suffer from it's own standard. Standards are to be used wisely and broken when needed.

Yes, but in the Windows world file permissions also do not exist, whereas in *NIX they do indeed exist which makes a virus's job impossible unless the person is dumb enough to run it as root.

Even if RPMs would "take over", you can use rpm2tgz in Gentoo in order to turn them into tgzs and then extract them. However, portage isn't in danger of going anywhere because it compiles stuff from the source, and seeing that having the source available is key in making "open source" software, the source shall stay.

..incorrect. they very much so do indeed exist. they're just less-known
because you don't have to screw with them half as much.
its a lesser known whichcraft. thats all

go into windows, right click any file. to properties, and then you'll see
the series of check boxes and whatnot. therein, lies your "file permissions"

:p

That's right. They're pretty much useless though since almost every windows user I know runs everything with admin privileges. Yeah, maybe windows is not to blame for that, but still...

[placeholder]

..incorrect. they very much so do indeed exist. they're just less-known
because you don't have to screw with them half as much.
its a lesser known whichcraft. thats all

go into windows, right click any file. to properties, and then you'll see
the series of check boxes and whatnot. therein, lies your "file permissions"

:p

See, I don't use Windows except at school and here are some things that I can do as a normal user.

1) Uninstall programs
2) Install programs
3) Delete important system programs and folders

That along with more. You cannot tell me that someone running as a normal user wouldn't be able to infect their computer, because you know that they would indeed be able to do that. In my long and tedious experience with Windows, the only "permissions" were read-only, hidden, and archive. That's where it stopped, and there were no permissions to assign them to users or anything. That was with WinXP last December using NTFS(supposedly their most advanced filesystem although I didn't notice a huge difference between it and FAT32) so I think that if more useful permissions existed then I would have seen them.

We already know that unless there's a huge bug that can setuid root to a malicious program, then nothing is going to hurt a Linux system. Also, MS encourages people to run as an admin, since that's the first account ever created which most people never change.

[episode96]

That along with more. You cannot tell me that someone running as a normal user wouldn't be able to infect their computer, because you know that they would indeed be able to do that. In my long and tedious experience with Windows, the only "permissions" were read-only, hidden, and archive. That's where it stopped, and there were no permissions to assign them to users or anything. That was with WinXP last December using NTFS(supposedly their most advanced filesystem although I didn't notice a huge difference between it and FAT32) so I think that if more useful permissions existed then I would have seen them.

Wrong. Windows 2k/XP NTFS security scheme is frankly quite advanced, you should check it out again.

Also, MS encourages people to run as an admin, since that's the first account ever created which most people never change.

Agreed. Windows installation should educate people about basic security and encourage the use of a normal user account, something which most major Linux distributions do.


Anyway, I don't think windows security problems hace actually much to do with the topic we're talking about. Standardizing Linux is surely a good thing IMHO

[virtual]

Hi,

File permissions do exist, but that was not my point.

Imagine you want to write a virus, a virus needs a host to breed and multiply.
What do we know, well let's say we use something called the registry and we put this on our pc, we'll make this file our gathering point for all our confiigurations and we will make it a standard so all pc's use this file.

Now if I can infect the registry I can infect "all" pc's because I know how this registry is built and it's the same on all pc's.

In Linux writing such a virus is a bit more difficult because there is a larger and diverse set of config files, and these are not even located on the same place on different distos. So even if the virus could infect it would not be able to breed as rapidly because of the non standard way things are done in the Linux world.

This is just food for thought, an example of a trap when using standards of course you can secure any os. Standards are a good thing but they should be open for re-evaluation constantly and not be so rigid that they become a ball and chain around the leg of progress.

[Saa]

In Linux writing such a virus is a bit more difficult because there is a larger and diverse set of config files, and these are not even located on the same place on different distos. So even if the virus could infect it would not be able to breed as rapidly because of the non standard way things are done in the Linux world.

That scenario is not taking into account permissions. Once again, an idiot would have to be hit by that virus to do the kinds of things windows worms and virii do right now. The virus would have to be running as root or a very potent user account for that kind of mess to take place. The reason a standard way of doing things hurts windows so much is because not only is everything the same, but 80% of the users are running with the same permissions.

[Evangelion]

not true since you can install binary packages on gentoo

Yes it is true. The original posted compared RPM's to Portage. That comparison is flawed since they are not the same thing. RPM is a method to package apps, Portage is a method of delivering apps to the system. More valid comparison would be RPM vs. Ebuild or up2date vs. Portage

I fail to see what binary-installation has to do with this.