Home > Linux > ISO Makes Linux Standards Base a Standard ISO Makes Linux Standards Base a Standard Thom Holwerda 2005-11-01 Linux 36 Comments The Open Source Business Conference here will open on Tuesday to the news from the Free Standards Group that the Linux Standards Base has been approved as an ISO standard. About The Author Thom Holwerda Follow me on Twitter @thomholwerda 36 Comments 2005-11-01 9:16 pm joelito_pr Now all we need is to keep adding more and more lsb compliant distros to the list (and ISVs actually making lsb compliant packages) Still, I’m a bit concerned about non-rpm based distros, I mean, to be fully lsb compliant, many need need to improve rpm support(natively) 2005-11-01 9:28 pm http://www.gentoo.org/news/en/gwn/20030401-newsletter.xml 2005-11-01 9:45 pm well 🙂 i don’t remember any other november fool’s joke 2005-11-01 10:29 pm joelito_pr LoL, that was a nice one(but still, we need proper RPM) 2005-11-01 10:48 pm It’s entirely possible to use Gentoo and creating a LSB compliant profile and produce RPMs. But who wants to restrict himself… 2005-11-01 11:19 pm A standard is a standard only when almost everyone is using it. I can go off and create an off-base Linux spec and call it the “Poo Standard”. No one is going to use it — but hey, I’ll call it a standard anyway! Pfft. 2005-11-02 12:48 am archiesteel Trolling again, are you? I guess you don’t know what ISO stands for: International Organization for Standardization. Here, enlighten yourself in between two trolls: http://en.wikipedia.org/wiki/International_Organization_for_Standar… BTW, your definition of standard is wrong: a standard is something people agree on, not necessarily something that lots of people use. Wikipedia defines a standard thusly: “In technical use, a standard is a concrete example of an item or a specification against which all others may be measured.” Are you that desperate for attention that you must turn every subject into a flamewar? 2005-11-02 1:28 am Wikipedia defines a standard thusly: “In technical use, a standard is a concrete example of an item or a specification against which all others may be measured.” That’ doesn’t mean is true, since when wikipedia is credicle? it has no credits nor is validated as a true knowledge institution. Ain’t nothing but an online wikie full of volunteers with doudfull credibility. 2005-11-02 1:34 am “since when wikipedia is credicle?” it’s as credible as anything written by humans. 2005-11-02 1:57 am archiesteel I’ll pass on the fact that your criticism was posted anonymously, which is quite ironic considering your charges against Wikipedia, or the fact that you made numerous spelling mistakes. If you disagree with the definition, you can register with Wikipedia and modify it. Since it’s peer-reviewed (i.e. based on consensus), if your definition is better it should be adopted. Wikipedia is a bit of a self-correcting encyclopedia. It may not always be accurate, but any inaccuracies have an increasingly higher chance of being corrected as time goes by. And since knowledge is based on consensus, it is ultimately as reliable as a printed encyclopedia (which are known to have errors). In any case, let me give you Merriam-Webster’s definition: “something established by authority, custom, or general consent as a model or example (CRITERION); something set up and established by authority as a rule for the measure of quantity, weight, extent, value, or quality” Hey, guess what? It’s practically identical to the Wikipedia definition! Now, did you actually have a point, or are you just Linux is Poo trolling anonymously? Because my original agument still stands, untouched… 2005-11-02 2:47 am Actually the meanings are quite different, since wikipedia talks about measures the other talks about rules, you see, and you are giving the reason saying that I can enter to wikipedia and change the definitin of standar, so I could enter and define standar as a “bunch of red oranges”, woul’d that make it true? I don’t think so, nothing backs up wikipedia definitions, only common sense, no cientific studies, nothing. 2005-11-02 5:37 am archiesteel Actually the meanings are quite different, since wikipedia talks about measures the other talks about rules, you see, I see, you don’t apparently: the Webster definition specifically mentions “a rule for the measures”. Read it again. “and you are giving the reason saying that I can enter to wikipedia and change the definitin of standar, so I could enter and define standar as a “bunch of red oranges”, woul’d that make it true?” Of course not. And it wouldn’t stay on the page either, because there are other users and moderators tracking edits. The indicent would be noted and you would likely get your user account revoked. So, if you’re asking whether Wikipedia is completely protected against this kind of pointless vandalism, then the answer is no. However, its contents do go through an informal validation process. In general, “eeer” user review plus the fact that a history is kept for all page tend to discourage this anti-social behavior. nothing backs up wikipedia definitions, only common sense, no cientific studies, nothing. You can’t generalize like this, you don’t know who wrote all 800,000 articles. Sure, the quality of information is uneven – in fact, some articles clearly indicate (in a box at the top of the page) that they need work. However, consider that a great number of articles are written by people who know what they’re talking about, since they’re interested in these subjects in the first place. Do you have any example of a page that contains erroneous data? 2005-11-02 11:29 pm Aren’t definitions axiomatic, I mean there is no scientific way to prove the definition of “number”. If there is, I would be curious how you could prove this “scientifically”. 2005-11-02 1:28 am Oh, I know what it means, and I know what a “standard” entails … But do you mind explaining to me of what use any standard is if it’s not being used by most of the major players? With Redhat, Mandrake, SuSE, Gentoo, Ubuntu, Slackware, and everyone else doing their own thing, what do they hope to accomplish by signing a nice little piece of paper that says “Okay this is an official standard now kthx”? 2005-11-02 2:11 am Who cares about all that? Think of all the red tape and marketing oppurtunities! On a more serious note, this is rather useless unless 1) Distros comply 2) Someone enforces the standard It’s definitely too early to say how much of that will happen, but we’ll see… We all know how badly POSIX worked. Someone has to start waving the big stick if they want this to work. On a side note, I can’t wait to see all the hypocracy from the anti-MS crowd. (.NET anyone? “covers only a small subset” BS? This is in exactly the same boat.) 2005-11-02 4:59 am Man … haven’t you learned? The Linux community is all about double-standards. To make an example of a few: Linux patches: “See how fast they patched it?! Open-source is great!” Microsoft patches: “Feh! Microsoft creates such insecure software! Another patch! What a POS!” Linux instability: “It’s because development is so rapid! It’ll get fixed soon!” Microsoft instability: “WINDOWS IS SHIT!1!!!! LUNIX FOREV4H!1!” Linux conference: *fap fap fap* Microsoft conference: “Boooo!” 2005-11-02 2:16 am archiesteel Oh, I know what it means, and I know what a “standard” entails … If you knew what it meant, why did you make a false claim? A bit too quick on the troll, were we? But do you mind explaining to me of what use any standard is if it’s not being used by most of the major players? With Redhat, Mandrake, SuSE, Gentoo, Ubuntu, Slackware, and everyone else doing their own thing, what do they hope to accomplish by signing a nice little piece of paper that says “Okay this is an official standard now kthx”? Before distros sign on to it, it has to be accepted as a real standard. The fact that it is now an ISO standard will be a powerful incentive. Mind you, most of the distros you named already have LSB packages that can be installed. In other words, you can already make a Mandriva install LSB-compliant (for example). Of course, it’s your right not to care. But I find it funny that you defend yourself from trolling on other threads when clearly that is all that you were doing here. A bit hard to play the innocent virgin when caught red-handed, innit? 2005-11-02 5:01 am Really? I can make a Mandriva installation LSB-compliant? Meaning it doesn’t come LSB-compliant out-of-the-box, hence eliminating any and all benefits to being LSB-compliant? Well that’s just f–king great! It’s what I’ve been wanting to hear! I have to install a distro, then take part in a grand hackathon just to make it LSB “compliant”. Seriously, you Linuxtards ought to take a damned step back and look at yourselves and what you’re saying. In any other culture at any other time about any other topic, you would be labelled mentally insane. 2005-11-02 5:21 am archiesteel Really? I can make a Mandriva installation LSB-compliant? Meaning it doesn’t come LSB-compliant out-of-the-box, hence eliminating any and all benefits to being LSB-compliant? I don’t know, I haven’t tried the latest Mandriva yet (I run it on my desktop – Ubuntu on the laptop). Why don’t you download a copy and try it instead of trying to start flamewars on web sites until you get banned? Well that’s just f–king great! It’s what I’ve been wanting to hear! I have to install a distro, then take part in a grand hackathon just to make it LSB “compliant” If by “grand hackaton” you mean “ticking a single checkbox during system installation”, then yes. 🙂 Seriously, you Linuxtards ought to take a damned step back and look at yourselves and what you’re saying. The only “tard” that needs to step back and look at what he’s saying here is you, Lippy. In any other culture at any other time about any other topic, you would be labelled mentally insane. That is probably one of the most pathetic insults I’ve heard on this site, and I’ve heard quite a lot. 2005-11-02 2:51 pm It wasn’t meant to be an insult — just a statement of fact. As for the Mandriva stuff … didn’t you just finish telling me that it’s an after-installation operation? 2005-11-02 4:57 pm archiesteel It wasn’t meant to be an insult — just a statement of fact. Yeah, right. Like me saying that you’re an idiot, right? I’m not insulting you, just stating a fact… Come on, trolls were using these logical fallacies on newsgroups and bulletin boards back in the 80s. Can’t you come up with anything original? I’m tired of responding to ad hominem attacks and strawman arguments…how about actually constructing a logically sound argument, for once? I know it’s harder, as it actually involves thinking about what you write, but it would be a refreshing change. As for the Mandriva stuff … didn’t you just finish telling me that it’s an after-installation operation? No I didn’t. I said they were packages to make the distro LSB compliant. These packages can be installed during the initial installation (by checking a box), or later if you want (by launching the Package Manager frontend, checking a box and clicking on the “Install” button). I you were anything else than a knee-jerking troll you’d have understood at. Not insulting you, troll, just stating a fact. 2005-11-02 5:24 pm Awesome. Well at least that’s good to hear. So we can take Mandriva off the list I suppose. Now what about the other 1199 distros? 2005-11-02 7:35 pm Ubuntu has an lsb-core(base?) package installed by default and you can add the full lsb framework using apt-get/synaptic 2005-11-02 12:26 am jdrake I have never seen any benefit to something like this. Better to support something like AppBundles that are versioned. Easier for all concerned. 2005-11-02 2:57 am joelito_pr That most comercial distributions will keep embracing the standard one way or another. However, I don’t count the many “kid in basement” type of distro, and it’s their right, but as a user, I’m glad that we have a linux development plataform we can agree with without vendor lock ins or weird license issues. 2005-11-02 4:09 am What I don’t get is why RPMS are being floated around as a possible “standard” for packages. What about debs? Given Debian’s history and superior dependency handling, which stretches back for years now, wouldn’t it make more sense to have deb packages be the “default standard?” Not that I use Debian. Trust me, I don’t. Were I ever to go back to binary distributions, my only choices would be Debian or Ubuntu, based mostly on the packaging format. I still can not understand why RPMS are even being considered. Perhaps someone can explain. 2005-11-02 8:30 am hobgoblin i think one would be best of if one didnt standardize on a package manger but rather on a lsb-compliant way of listing dependencys. one way could be to use a tar-ball and include a ascii file listing of the dependencys. then the distros own tools could read said ball+file and create a new package based on it for internal use. hell, one may even allow for installation of third party drivers this way by having a lsb way of automating the compile of it towards the current kernel. trow it at the distros package manager. it reads the included manifest and finds that something needs a compile, and then fires up the compile in the background. yes, security may be an issue. but anything can be. proper review is one of the things that a admin have to do. and for a home pc, the user is the admin, like it or not… 2005-11-02 1:01 pm …wouldn’t it make more sense to have deb packages be the “default standard?” This is mentioned in LSB FAQ: http://www.linuxbase.org/modules.php?name=FAQ&myfaq=yes&id_cat=11&c… For me this makes sense: other distros don’t support .deb, but Debian supports .rpm. Note that I’m a Debian user and enthusiastic, and I still accept this. A standard needed to define package format, and the committee selected rpm. No big problems here. I agree with you that .debs (or more specifically apt) has taken care of dependencies longer than .rpm distros, but that has changed now that most distos have some kind of automatic system (urpmi, etc) that fixes this. Eleknader 2005-11-02 7:29 am About bloody time. Now let’s hash out any problems and get this thing moving. 2005-11-02 2:18 pm TomHu Does an rpm package contain dependency information itself, or is it just those high-level tools (urpmi…) that maintain their own dependency list for each known package? The choice of RPM as the basis for the package format was made based on practicalness. Nearly every distribution has the ability to install a RPM format package, including Debian (using alien). As nice as the Debian packaging system is, other distributions just don’t have the tools to be able to process packages in that format. Seriously poor reason to enforce rpm compat. That’s not a standard it’s just:“Everybody does it. yes it’s poor, but still… everybody does it, so we enforce it.” One year ago I had the pleasure to meet “urpmi”, it didn’t work the way documented, and it didn’t do this reliably, thus I never believed it to be a way out of rpm-hell, which I left about 5 years ago. Poor technical design and implementation doesn’t make up for wide usage, see Windows, I thought the OpenSource community was about quality software. LSB also receives some criticism: http://www.livejournal.com/users/udrepper/8511.html I think enforcing an incomplete flawed standard-proposal as a standard is no good. However this seems to be a business decision about making money with “standards compliance” banner on boxes and support contracts. It simply sucks. 2005-11-02 3:34 pm Rahul “Does an rpm package contain dependency information itself, or is it just those high-level tools (urpmi…) that maintain their own dependency list for each known package? ” RPM and DEB package formats both maintain dependency information within the package specification RPM as a command is somewhat equiavalent to dpkg in Debian. They list dependencies but do not resolve them automatically. Yum in Fedora is similar to Apt-get in Debian. They both resolve dependencies automatically Seriously guys, the format is different from the tool. RPM (a subset) as a package format is what is specified for interoperability as a LSB standard Ulrich Drepper’s (Red Hat developer and maintainer of glibc) blog which you refer to is a analysis of LSB testing and how it should be improved. These have been acknowledged and improvements would be made 2005-11-02 4:25 pm “RPM (a subset) as a package format is what is specified for interoperability as a LSB standard.” Which to me is ridiculous. In order for a source distribution, such as Gentoo, to be LSB compliant, it must require the installation of rpm. This is nonsense. rpm2targz is small, works well enough and does not require rpm be installed. One can only hope USE=”-lsb” will be available in such distributions. 2005-11-02 5:02 pm archiesteel I don’t know what the big deal is, I use both RPMs and Debian packages (my desktop’s Mandriva, my laptop’s Kubuntu) and to me they are equally practical. I really don’t feel on is superior to the other… Synaptics does feel much faster than rpmdrake, but I believe that’s because Synaptics is written in C (or C++) while rpmdrake is written in Python (or another similar interpreted language). In any case, it’s clear that the LSB had to pick one, and the reason they gave is satisfactory in my view. 2005-11-02 7:39 pm One thing I have noticed between at least Apt vs Yum, (I’m not sure if it’s down to RPM vs. DEB or not) is that Yum is SLOW. I have four computers that I use pretty much every day. 1) Desktop w/ AMD64 3200+ 1gb Mem, running Ubuntu 5.10 2) Laptop w/ Pentium M 2ghz 1gb Mem, running Ubuntu 5.10 3) Server w/ Dual 466mhz 384mb Mem, running Debian Sarge 4) HTPC w/ 2.26ghz 512mb Mem, Fedora Core 4 Even #3 with an ancient Dual 466mhz processor in it will install packages faster than Yum does. The problem I have with Yum, is that it seems to update the repositories every time you try to install something. Whereas with Apt you can just state ‘apt-get update’ if you’ve already updated your package lists once for the day, why would you need to update it again 10 minutes later? Some of the RPM based distros are quite nice, but their package management leaves a bad taste in my mouth. The only reason I went with Fedora Core 4 on my HTPC is because for some strange reason, the newest version of Mythgame is not packaged for Debian. I still think Urpmi works better than Yum. I haven’t ever used any previous Fedora Core releases, so I don’t know if they broke something with v4 or not, but it just seems to take forever just to update some packages. 2005-11-02 10:06 pm Rahul “One thing I have noticed between at least Apt vs Yum, (I’m not sure if it’s down to RPM vs. DEB or not) is that Yum is SLOW. I have four computers that I use pretty much every day.” It works that way because it checks the repository each time instead of requiring you to run apt-get update manually. You might find the yum -C switch to be quite useful here. see the man page for details 2005-11-02 11:58 pm “It works that way because it checks the repository each time instead of requiring you to run apt-get update manually. You might find the yum -C switch to be quite useful here. see the man page for details” Thanks, I’ll have to remember that. I wonder if there would be an easy way to do a benchmark between the speed of RPMs vs Debs. If I recall correctly, one of the supposed advantages of RPM over Deb was that it used gpg, which they finally added to the 6.x version of apt. But I still think that RPM does it per package, whereas apt does it just on the repository? not sure about that, but it could be why rpms take longer to install.