KDE Installer Project 111
An Anonymous Coward writes: "There have been countless requests from KDE users, on the dot, on the lists, and even elsewhere, for a KDE Installer and Updater. Nick Betcher (aka Error403) has stepped up to the challenge and now needs your help to make this project really happen. His current code is in CVS and the project is in active development. The install starts off with an intro/detection screen, prompts the user for the type of installation, prompts for the destination of the KDE installation, and then prompts for the packages to install (see all the screenshots)."
Re:installer (Score:1)
Listen. Some people use their computers to do things. Some people use their computers as toys. Those of us who actually want to do things other than constantly upgrade it don't have the time or brain capacity to memorize dozens upon dozens of strange heiroglyphics(sp?).
As for the argument that typing the command is easy, OF COURSE IT IS. By this reasoning, everything about computers is easy. Most of it is just typing, and if you have the move it, even the monitor doesn't weigh over 50lbs. Hell, I used to run a moving van for a living. Carrying a solid-oak triple dresser is hard. Picking up a loaded filing cabinet by yourself and moving it ten-feet is hard. Typing a one line command is damned easy in comparison. Remembering that command when you only use it once every 6 months is a challenge though.
Re:This is a great move! (Score:1)
Um, that's as silly as saying "Go get Debian, and type apt-get update; apt-get upgrade". The kind of folks who would feel the need for a KDE installer really wouldn't be installing another OS on a lark.
Re:Package Managers and Installers (Score:1)
Unfortunately, RedHat/Mandrake/Helix^H^H^H^H^HXimian/etc... and their "NIH" syndrome has caused each one to write continously *bad* implementations of tools that incorporate exactly the functionality that APT already has, rather than take APT as a "middle layer" library, tell it to use 'rpm', 'pkgtool', or 'dpkg' at the package level, and build a front-end to it that people will like.
This is the Unix philosophy, folks! Small programs/libraries that DO ONE THING *WELL*, chained together in such a way to perform complicated tasks, instead of big bloated monstrosities that do ten thousand things poorly (or ten thousand programs that do ten things poorly, as is the case with the package tools). rpm/pkgtool/dpkg are all perfect tools for installing and deinstalling packages. APT is the perfect tool for managing a large library of package sources and selecting/downloading and then installing (using the package tools native to the OS) packages based on dependencies. dselect/apt-get/gnome-apt and whatever front-ends the future holds need only be concerned with the actual user interface, not the intricacies of package management or installation.
BTW, to forestall any flames, I *have* used almost every installer/distro/updater/freenix out there at some point, and IMNSHO *all* of them have major flaws. Even debian, but the flaws I have troubles with are being solved rapidly, and it is shaping up to really stand head-and shoulders over everyone else.
Re:KDE installer will get many more users... (Score:1)
Re:Issues I need to discuss (Score:1)
You should probably calm down a little, telling people who don't like the idea to go away, even though it might be a reasonable sentiment, isn't a good way to get contributors. The best thing to do is probably to ignore flamers, they'll tire soon enough and go away by themselves, unless they actually have good points.
Re:A must-have for KDE (Score:1)
Re:Why? (Score:1)
If you were using Debian, you would just add the path to your sources.list and do `apt-get update ; apt-get install [package]`, and it'll automatically download any dependencies.
Re:Why? (Score:1)
No, there doesn't seem to be an option for dpkg or apt-get to retrieve an package from FTP/HTTP URL which you specify, but I really don't think it's such an important feature.
Re:Update! (Score:1)
You have to understand that most of the people at Slashdot these days are TROLLS who can't live and let live. So he's getting a lot of e-mails from people saying "An installer is a STUPID IDEA, and YOU FUCKING SUCK! GO KILL YOURSELF!" and he has to wade through all that crap to get to the suggestions that actual potential users of his program have to offer him.
He's probably never done this before, and he hasn't generated the thick skin that he needs. If he has talent, he'll develop the thick skin quite quickly. But right now, it's depressing him and pissing him off, and he wants the insults and lameness to stop. What about that makes him an egoist? He's tired of being put down, and he's being BLASTED by debian people who DESPERATELY want to believe that nothing more than apt-get is needed. The problem is that even though a baby could logically use apt with no problem, people DON'T ACT LOGICALLY. They want GRAPHICS to make it easy, not a command line where they have to think. Apt-get is too involved for these people. They just want a button that says "GET KDE!".
Constructive Criticism does NOT include people saying "YOU SUCK!" Listening to comments from people who don't like the idea of his program is not constructive, it's wasteful. If you say that lynx has everything in it that Mozilla doesn't have, then I'm just going to tell you to use lynx. In the meantime, that has done NOTHING to improve the quality of my product.
Just give this guy a break, ok? He's stumbled upon a religious issue that he didn't know existed.
Re:Helix installer and KDE (Score:1)
Re:aduva (Score:1)
Package Managers and Installers (Score:1)
I believe that installation of packages is the job of the operating system. So, under Linux, it'll probably be am RPM, DEB or tar.gz file. Under Solaris and Unixware it'll be the standard package format. I'm not sure about other Unices.
What's needed is a standard interface for installing applications, checking for updates, and where appropriate, downloading new versions. At the lowest level is the package format. On top of that should be an abstraction layer that interfaces with the operating system command. For example, a call to "InstallApp()" would use the "pkgadd" command in Solaris, but "rpm -i" in Red Hat. Other functions could include RemoveApp(), CheckDependency(), LocateUpdate() etc.
Finally, on top of that would be the application installer, which would hopefully be part of the OS (such as SCOadmin, Linuxconf, Admintool, SAM, SMIT etc), or a perhaps a custom installer such as Red Carpet, or the KDE Installer. In this case, the custom installers would essentially be a subset of the main OS-level installer.
Now, someone here will probably respond with the typical "if you want it, write it yourself", but consider the co-ordination necessary for such a project, partly due to technical differences between the various packages, but mainly due to the egos of some developers that insist on reinventing the wheel the "proper" way.
But something needs to be done, because they way things are headed, there will be a plethora of incompatible installers, designed to promote a particular application or project. This will only confuse users and make life more difficult for people migrating to Linux (or Unix).
Re:Why? (Score:1)
siri
Re:KDE Installer, upgrader? (Score:1)
Yes. You forgot dselect update select and apt-get dselect-upgrade. :)
--------
Genius dies of the same blow that destroys liberty.
Help optimize dpkg/apt (Score:1)
Rather than writing InstallShield for Linux <GAG!>, we need help rewriting dpkg, because right now it's a little inefficient and it doesn't support some things very well (like totally unattended installs).
The last thing we need is an essential system component tied to a GUI. That is simply not the UNIX way; it's the Windows wa.
--------
Genius dies of the same blow that destroys liberty.
Re:What this installer doesnt do. (Score:1)
Maldivian said: "Your main auidence are the newbies. There is no need to say all that crap to them."
Learn to read carefully
Re:Issues I need to discuss (Score:1)
Oh please... he's just fed up with some of the comments here and it shows - Give him a break!
BTW: This installer is about KDE, not Linux. Since KDE is NOT a Linux-only product, all comments of the form use rpm/dpkg/apt... are strictly speaking off-topic
Re:Ahum (Score:1)
> where Nick himself posted. Incidentally, one of
> them was modded down to -1 (does it signify
> something when the originator of the article is
> moderated "offtopic" and "troll"?), so you'll
> kinda have to hunt for it.
Well, it does tell a lot about the moderaters.
Johan V.
Re:This is a great move! (Score:1)
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:A decent GUI front end to APT (Score:1)
And I agree that whoever came up with that kpackage error message should be dragged out into the street and beaten by someone yelling "ERROR MESSAGES MUST NOT SUCK!!!!!"
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:Why? (Score:1)
Re:No you don't. APT does it, is tested, and works (Score:1)
The Crashtester list hosted by Mandrake has Mandrake employees calling the next version 7.3. Either way, it probably doesn't matter. And yes, APT would be worth a major revision, IMHO.
Re:Update! (Score:1)
Mandrake to use apt? (Score:1)
THe recent RPM port of APT (it was originally designed to be portable) by Connective means that Connectiva now uses this system and Linux Mandrake 7.3 will in the future.
Really? That's a bold move; if it works, it would really move them out of Red Hat's shadow. apt-get is such a wonderful tool, but I wonder if their packages and dependencies will be constructed well enough for it to work like it does on Debian.
Unfortunately, based on the hash they made of update-menus, I'm not overly optimistic.
but! (Score:1)
Re:Why do we need standardization? (Score:1)
However, were it not for the POSIX standard, W3C standards, OSI standards, IEEE standards, we probably wouldn't have gone this far by now.
It all summarizes (again) to the "tradeoff" word - are there many innovation possibilities there? If so, make no standard yet. If not, a standard might be a good thing.
Hmmm. (Score:1)
Is there one who has one yet?
Hard? (Score:1)
17 (Score:1)
Possibly because they don't care if a good net installer makes yo sell less distro boxes. And they didn't need to come up with a payment model to replace the unsold boxes.
different strokes for diffrent folks (Score:1)
Re:KDE needs smaller packages (Score:1)
Re:KDE needs smaller packages (Score:1)
Re:Just don't reinvent the wheel (Score:1)
I can gaurantee you that this is not the case. KDE developers have expressed over and over that they want the distributions to deal with the packaging problems. KDE only makes two types of packages, Source and Debs. The rest are sent back to KDE from the distros before release. This guy decided that because some distributions lacked an easy way to install KDE that he would try to make the problem better and I praise him for this.
 
Matt Newell
Re:Why do we need standardization? (Score:1)
signature smigmature
Re:This is a great move! (Score:1)
signature smigmature
Re:KDE Installer, upgrader? (Score:1)
Re:KDE bows to user demands. (Score:1)
I don't want all of my programs each having their own installer with their own UI and all the usage quirks that go along with it.
Something like that sounds like the myriad of windows installer programs like installshield.
I want one monolithic packaging tool that handles every package.
--
Re:KDE bows to user demands. (Score:1)
It's called APT and it comes with Debian [debian.org].
--
Re:We don't need stinking installers! (Score:1)
A must-have for KDE (Score:1)
However, I think the program eventually needs to appeal to all but the "I want to compile the source" people. This means very customized install options, as well as a default newbie option. Individual packages, while not supported yet according to Nick, are also essential to its usefulness.
In addition, such a (hopefully) broad install program should support easy and clean uninstallation of KDE packages. I recognize the availability of kpackage to manage RPMS, but this program would potentially download the binaries of packages required to install (like Ximian GNOME).
I kind of went dreaming there for a sec, asking a lot of a new program. In short, it looks like a great start, and I encourage those interested to help KDE in a sore area.
Re:KDE Installer, upgrader? (Score:1)
Install from source! (Score:1)
Many times a new version of some package comes out that I want to try, and it's only available as source code. What we need is a "universal" installer that can build a whole system or any of its parts from scratch, starting with downloading the source code.
Actually I already started a project [lfsmake.org] like this, using Makefiles. Someday I'll find the time to get back to it....
Re:This is a great move! (Score:1)
over-night job for my 400 MHz machine.
Also, you need to compile things in the right
order, and you need to run these commands for
each package seperately - which makes it a drag
for a newbie.
If you have a compile script for all the KDE
packages and Qt it's easier, and comes down to
waiting, but unfortunately KDE doesn't provide
one.
There are some additional complications, too -
like compiling konqueror (you need to have
various other libraries installed to have all
features) and i18n packages seem to want to
default to
/opt/kde2
Having a good compile/install script which is
likely to work on most platforms would be a
worthwhile project, too..
There is also the problem that if the install
fails somewhere in the middle you now have a
broken desktop. (For example lets say a package
didn't download fully.) Then the user will
have a hard time fixing it (e.g. downloading
the package, but not having konqueror anymore).
These things are easy to solve for most of us,
but I think it's worthwhile for KDE to offer
an easier way to upgrade.
Re:Why do we need standardization? (Score:1)
Standards should be carefully designed so that (1) they don't need to be changed later and (2) are extendable.
I hope this doesn't seem contradictory. When a standard is designed, the designers should try to establish a reasonable base that they don't think will change much, and also leave room for things to be added, or even modified later.
I'm thinking specifically of HTML. Ignore the fact that browsers and HTML coders alike are very liberal with the standards. HTML that conforms to standards has the version number embedded in it. A standard browser should be able to read the HTML and render it accordingly.
If the browser is newer than the HTML, then it simply has to read the version of HTML and use the standards for that version to render the page (point 2 above). If the HTML is newer than the browser, then we hope the newer HTML isn't too different from the old (point 1 above).
Did ya hafta FP? (Score:1)
i should change my sig to "fsck karma"
"Me Ted"
Re:Did ya hafta FP? (Score:1)
"Me Ted"
Re:Why? (Score:1)
-Nano
redcarpet? (Score:1)
Re:Flamebait Troll (Score:1)
Re:A decent GUI front end to APT (Score:1)
Great Idea! Actually gnome-apt already does some of this, but it is still in heavy development, plus that some of it's behaviour is utterly braindead (like remembering what columns I don't want to see in the package display, but continually forgetting that I want my packages sorted by section, not alphabetically).
I have however seen that gnome-apt was not upgraded when I went from potato to woody, so does this mean that Debian is discontinuing it? That would be a shame, as it would seem like a good base to build this project on.
Martso long as ... (Score:1)
Re:A decent GUI front end to APT (Score:1)
Re:Hard? (Score:1)
Open Source development/responsibility (Score:1)
I see here a potential problem in the open source philosophy. if this installer is so badly needed, who is going to finally decide what it's going to look and feel like? And who benefits from that decision, especially if something really innovative happens with it.
Nick Betcher, of course, some of u are going to answer. But the whole point of him asking for this input is to get collaboration from other developers.
Linus did something really revolutionary with the help of thousands of developers, and it sent him into demi-godhood. Is that fair? Not everybody wants to hide in the shadows of some charismatic lead software architect. Furthermore, tho nobody can profit directly off the code, there may be profits from the building of packaging and support services for given software. Who has first crack at that project?
That said, good luck, and I wish you all well with these tough choices.
Re:What this installer doesnt do. (Score:1)
I was also responding to his earlier comment [slashdot.org] :
* IF YOU DONT LIKE MY IDEA OR PROGRAM ITSELF GO AWAY! I only want constructive critism. All lamers flaming me will be ignored and no brain power will be put into reading your lame @#$ comments
In light of that, he seem to want only praises about his software from newbies. What's his audience?
Re:KDE bows to user demands. (Score:2)
1. It makes sence for this to be an RPM only utility.
2. For those distributions with nothing similar, it can become that universal tool.
Sure Debian has this one feature we all like. However there are other things about Debian I don't like so what do we do ?
Either convince Debian to change the way it dose things to suite those who don't even use it as is. ( fat chance ). Or we simulate APT on other distributions.
Snideness aside, isn't this a good idea?... (Score:2)
Re:Oh NO!!! (Score:2)
While everything you say may well be true, I think you may be over judgemental on a few points, such as:
the guy who is "writting" it is a very new programmer with unfortuantly a very young attitudeOh, so just because someone is young and nieve doesn't mean they might have a good idea? I mean, who wants a web browser that supports inline images? And Napster can't be any good - I mean, how stupid could someone be to write a really simple piece of code just to trade music files. Everyone knows the music industry will put an end to that - he was so immature to even try. And how, of all the rediculous things, how could someone leave a good education at Harvard to go write software? (Hey, you may not like him, but you got to admit that they way they started MS when they did took guts.)
it has no planning and the only clear goal is to create an installerSo? Blue Sky Software set out with that lone goal many years ago. They were so successful that they had to change their corporate name to that of their flagship product: InstallShield.
I do realise that KDE needs an installer but unfortualtly is a bit of a way off yet.Well, rather than whining about it, why don't you go write one? Sound's like that's exactly what this guy decided. The way I read the /. post, it sounded like the project was just starting, which is why they (he) was soliciting help.
You know, I just won't go into the whole apt thing right now. Granted, it's a great installer, I just wish it ran on Mandrake (which I use because it was the easiest to install at the time). I will note that it's a little strange that the best way to install KDE is with apt when not too long ago the Debian folks wouldn't touch KDE with a ten foot pole and that's partially what precipitated the start of Gnome. . .
-"Zow"
Re:This is a great move! (Score:2)
an easier way to upgrade.
But I doubt that the installer is it. You'll have one easy way to do it with Redhat, another for SuSE, another for Slackware, another for Debian, another for FreeBSD, another for OpenBSD, another for Solaris, another for AIX, another for IRIX...
Maybe if the Linux crowd got involved in the openports project...
Re:This is a great move! (Score:2)
Cheers!
Re:KDE installer will get many more users... (Score:2)
Yes it has... it's called MandrakeUpdate, or apt-get, or any of a dozen other distro specific apps.
Although no one within KDE is saying "stop", the KDE team has, over and over again, said that it will NOT make an installer - in fact the KDE team dosen't release packages... for a Good Reason: they see packaging and installation as a distro's job. That way, the distro can choose your own flavor of packaging (deb, RPM, tarfile), your own path system (/usr/bin, /usr/local/bin, /opt) and choice in systems is encouraged.
As a philosophy, this makes a helluva lot of sense: Slackware and Red Hat are on even footing when it comes to releasing KDE's latest and greatest. And if the new IBMinstall uses ".iop" files, then they can be easily packaged to that. Also, to address the monolithic nature of the structure of KDE, except for a few, well documented outside dependancies, KDE satifies it's own dependancies at compile and runtime. Most KDE apps are located outside of the KDE project as a result (or at least are not in the core groups: kdecore, kdegraphics, etc.) Many more can be found at: apps.kde.org [kde.org], a freshmeat like listing.
All in all, the philosophies of KDE are well thought out, and make sense (IMO)... it just takes reading or talking to a few developers.
--
Evan
Re:Why do we need standardization? (Score:2)
For software this is no different. If you look at the win32 platform for instance, you'll see my point. MS has succeeded in getting people to use their components. This has resulted in a large number of applications. From a technical persective there's nothing revolutionary about their stuff (some might even claim it is buggy and has obvious flaws). The Java platform is another example, by providing the same set of components on each VM, applications can be WORA. Often you don't even have to test it works and you can just assume it works.
Now under linux there's always more than one solution to a single problem. Whether we are talking about package management, window managers or even glibc. You can't make much assumptions about what will actually be running on a random linux install. This is both an advantage and a disadvantage: you get choice but at the same time, application developers must assume the common denominator rather than the latest and greatest to be present.
Re:No you don't. APT does it, is tested, and works (Score:2)
Re:This is a great move! (Score:2)
rpm --rebuild
> Sure it takes longer to build a program than it does to install it. But it ain't much harder (./configure; make; make install). And you get to be 100% in control.
vi package.spec
rpm -ba package.spec
There are loads of advantages to rpm. If that makes me anally retentive, so be it. That's not a bad quality if you're a sysadmin.
For your machines at home, please do as you see fit. For machines that need to be administrated, rpm is my choice, and I have the feeling I have more control than them compile-from-source people.
--
Re:What this installer doesnt do. (Score:2)
First of all, there's a really useful word called "you're". Try using it sometime instead of "your". Only when appropriate, of course. Errors like that, especially in the screenshots, make your project look really unprofessional.
The Debian references did come up rather often in these comments, but what you said comes off like a flame against Debian. The remark that it "isn't as cool" is especially glaring. If you didn't mean to sound angry, why did you use phrases like "little pet apt"?
"If [you're] not a newbie, or you dont like this idea, please dont talk about it"? So you only want to hear from people who like your project exactly the way it is? Constructive criticism is essential if your project is going to succeed. If you tune out all criticism you're probably going to release something which nobody likes.
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:Issues I need to discuss (Score:2)
Anyway, you're not going to gain much popularity by saying, "IF YOU DONT LIKE MY IDEA OR PROGRAM ITSELF GO AWAY!" You go on to say you only want constructive criticism, but it would appear that criticism, constructive or not, is the one thing you don't want. At first I thought you seriously had something to offer Linux with this project, but now I see you just want people to worship you over this installer.
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:Update! (Score:2)
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:Issues I need to discuss (Score:2)
The comments he seems the most fed up about are the ones that point out that he should base his installer on some existing installation system, instead of reinventing the wheel. His response is that he has spent "far too much time on this to start over". Translation: "No, this is MY installer! MINE MINE MINE!"
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
Re:Force install (Score:2)
Thanks for the caps.
I determined the information myself, but confirnmed it with Maceij and George from Eazel when they were in Australia for Linux.conf.au. They're not Ximian employees, but they certainly know GNOME well enough and are familiar with the Ximian installer.
Re:Why do we need standardization? (Score:2)
I'd rather have most everything work one way and have a few quirks than have a half-dozen quirky methods.
KDE needs smaller packages (Score:2)
Just don't reinvent the wheel (Score:2)
This sounds like kde wants to do it's own package management, which is scary. Package management should be easier, yes, but kde is simply the wrong place to fix the problem. Agreed, rpm isn't enough. Several dists provide the extra utils needed, the shining example is apt. Locating and installing packages is a breeze -- once you've configured the repositories you can just run apt-get install task-kde.
Now if you're talking about a gui layer above the OS package management, excellent! Debian has an excellent system in place, i here Mandrake has it's own too, and there are a number of upstart projects (check freshmeat). Just don't build it directly above rpm, there's no need to reinvent the wheel.
I haven't used an rpm based distribution in years, so i do have on question about rpm. Debian has a nice system for configuring packages (debconf), you can configure it to ask only a certain level of questions (critical,high,medium,low), remember your answers, and use a variety of interfaces. It'd be easy to add kde support to debconf, but how about rpm?
KDE Installer, upgrader? (Score:2)
apt-get install task-kde
and
apt-get upgrade
need I say more?
And unlike a 'KDE installer' or a 'Ximian installer' this functionality isn't limited to one piece of software. Check out Debian [debian.org] or Progeny [progeny.com].
Re:Why? (Score:2)
kdesupport needs OpenSSL. Okay, go and fetch that from rpmfind.net. Try again. kdelibs needs libmng etc etc. RPM only informs you of the dependencies, it doesn't resolve them. You also need to install in this order:
kdesupport, kdelibs, kdebase, then the rest. KDE 1 came with a very nice little install script, but that wouldn't work now that KDE is installed by default in different places by different distros (it wasn't in RedHat 5.2 when I first discovered KDE). The Ximian installer is nice, but according to a poster above, it force-installs everything, which has potentially disastrous results for your desktop, since it's not just GNOME rpms you need but stuff like libxml. apt-get sounds great, but Debian itself is for l33t3r h4x0rs than me.
Re:This is a great move! (Score:2)
Re:KDE Installer, upgrader? (Score:2)
What an amazing piece of software! (Score:2)
Somehow I think that this 'developer' doesn't know his arse from his elbow.
Re:What this installer doesnt do. (Score:2)
Sorry, but I'm of the opinion that you simply don't have a clue.
Website (Score:2)
This is a great move! (Score:2)
I tried upgrading to KDE 2.0 from a Mandrake 7.1-based install, and it wreaked hell with the login manager (which was based on KDE). Whenever I tried to quit X, it booted me back to the broken graphical login screen. It was more bug-infested than Ultima IX, so bad I couldn't even launch the file manager without it crashing. I followed the exact instructions on the KDE web page, and a small note (buried in the ftp folders) told me that I needed the latest version of menu and deskdrake
menu? What's 'menu'? Does it mean menudrake? I'm confused.
I tried to get help on the #linux channel on EFnet, but of course I was rebounded to linuxdoc.org, which made no mention of the problems I was having.
I ended up having to do a complete reformat and reinstall of Mandrake 7.1. Very user _unfriendly_. Helix Gnome's installer, on the other hand, was terrific. I haven't had a problem with Gnome since.
Re:Amiga had this in .... 1990 (Score:2)
The thing that Amiga had, which made it *really* pain-free installing, was a standardised directory structure, and an extensible but backwards compatible shared library system. It had this in 1985.
Not KDE's job! (Score:2)
Feeding a previously d/led rpm to a package manager doesnt cut it. It ought to act like a catalog of software that you can pick stuff you want and it'll D/L the RPMs and handle dependencies. Debian does this, but not in a newbie-friendly way.
Re:What this installer doesnt do. (Score:2)
By alienating users, you would never get anyone to support your projects.
1. Your comments towards Debian are totally unnecessary.
2. Your main auidence are the newbies. There is no need to say all that crap to them.
3. Attitudes need to be restricted. Ego's need to be --purged. No one will help develop your software if your at the lead of that team. I know I would never want to join your team.
4. Be polite.
Re:Why do we need standardization? (Score:3)
Because the formats are open, old formats can be converted to new.
Consider Linux binaries. The basic format is pretty standardized. Is that cutting off innovation? Unnecessary differences, on the other hand, are an incredible obstacle to progress. Consider how many programmer lifetimes are wasted dealing with byte-swapping, or because Unix, Mac, and PC ASCII files use different end-of-line codes. And there's no real advantage to any one of these.
I think the time has come that we've learned enough about what is good enough and what isn't, and need to try to merge a lot of these efforts into one great installer, rather than a bunch of limited ones.
KDE bows to user demands. (Score:3)
Someone has finally caved in and written an app which is functional in it's 1st public release. Meaning he probably didn't start last weak.
Now the trick is to get distributions to adopt this and extend it for none KDE apps and tune it to the specific distribution. It's not like it's from a company that can sell services around it as is and it doesn't make sense for free projects to spend too much time on the unnecessary incompatibilities between distributions.
The Slashdot Negativity effect. (Score:3)
"If your not a newbie, or you don't like this idea, please dont talk about it."
and calling him *egocentric*.
First of all, Slashdot has been completely infiltrated by people who want to do NOTHING but troll, piss people off, and do nothing constructive. If I posted a question to Slashdot, I would have to be prepared to ignore the majority of the posts as they would serve no constructive purpose whatsoever, and would just be attempting to debase the entire idea. This place is literally collapsing because of the amount of CRAP that people spew.
Now, this guy wants to make an installer FOR NEWBIES THAT WANT AN INSTALLER. That
is ALL he is writing this program for. I'll never use it. Many people here will never use it, but
maybe my dad or mom or other new Linux users will use it. However, people here can't
seem to get the concept of live and let live run through their heads. That's why we have a
constant bashing of Mozilla each time a new release comes out, by people who REALLY want to run lynx. So he's going to get all kinds of CRAP and INCENDARY suggestions from people who really doesn't want anything like what he's going to produce in the first place. This guy just doesn't want to have to go through MOUNTAINS OF CRAP to find the few suggestions that his REAL users want to give him. No, not EVERYONE wants to use apt, people these days have a complete fear of the command line. No, he doesn't want to use Redcarpet, can't we just leave it at that?
Now I know that everything here is going to turn into a HUGE flamefest, but PLEASE don't forward your comments to his e-mail or whatever UNLESS you really want to help.
Oh, and as for the lame grammar corrections, I think it is clear that this person speaks English as a second language. Not all programmers can speak english perfectly. I can understand what he is saying, even with incorrect grammar.
Update! (Score:3)
UPDATE: It turns out that Nick Betcher is an insecure egotist with nothing to offer the Linux community, trying to pass off a web page with crappy screenshots as a "project".
If you don't believe me, read one of the threads where Nick himself posted. Incidentally, one of them was modded down to -1 (does it signify something when the originator of the article is moderated "offtopic" and "troll"?), so you'll kinda have to hunt for it.
--
Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
No you don't. APT does it, is tested, and works. (Score:3)
No offence, but you most definitely don't. The Helixcode installer force installs every package. This is a bad thing, and even worse that Helixcode choose not to document the fact broadly. Thus IRC channels and mailing lists are being perpetually flooded with users complaining the installer has raped their system. Try jumping on mandrake-expert sometime. It ain't pretty sorting out the mess. Ditto all the Red Hat 6.2 users who installed Helix before trying updating to 7.0
You could use Red Carpet, but frankly I don't think much of Ximian or their products (most notable the registration of about ten or fifteen KDE tradmarks on Google adwords).
Their already a packaging system independent automatic installer avaliable, that is tested and proved. Debian has used it for a long time, and many people use the distribution simply because of this (Debian is too hands on for me, I like the automation of other distributions). THe recent RPM port of APT (it was originally designed to be portable) by Connective means that Connectiva now uses this system and Linux Mandrake 7.3 will in the future.
KDE installer will get many more users... (Score:3)
Integration with APT, RPM etc (Score:3)
Basically, any effort that goes into a GUI installer should be able to be utilised by apt (and rpm equivalents) without too much trouble -- they should not depend on having a GUI, and should not depend on any particular package format or file locations... it's all too complex :(
rr
Why do we need standardization? (Score:3)
Suppose someone develops a new, even better installer. But since it's not the same as your hypothetical Ximian/Helix universal installer, it would be shunned for not meeting the standard. Standards are made to be broken. It's hard to push the envelope in terms of technology and features when you're stuck with a strict standard to obey. Some of the best software programs have been the ones that have "thought outside the box" and rebuilt everything we know about a particular type of application. (Should we fault Linux for not supporting the "Windows standard?" I think not.) So let's not worry about setting a standard -- the best techonology will always set its own standard.
There's always different tools for different tasks. RPM might fit one project better, while the Debian packages might be the best choice for a second project. Let's not turn a hammer and a screwdriver into the same thing just for simplicity's sake.
Yu Suzuki
Now for a installer for everything else... (Score:3)
C:\
C:\Dos
C:\dos\run
installer (Score:3)
apt-get update;apt-get -yu dselect-upgrade
not only does that upgrade kde, it upgrades all of my system, and yes of course it can be set to only upgrade one package..
"I hate having to download 25+ rpms/source files and having to figure out libraries etc" Funny, since with Debian none of this is needed! Sorry if this sounds like another "apt is great, debian is great", but it IS. If you don't like having to install KDE on redhat with many odd rpms, then move to a decent distro based on
What this installer doesnt do. (Score:3)
* This installer doesnt manage any packages. I dont know where you guys got that from, its false.
* This installer doesnt replace all the Debian users' little pet apt. If you are using Debian, please dont mention that all you need to do is a "apt-get blah blah blah". WE KNOW. And if your asking yourself "Why not just use Debian" its because apt doesnt have a cool installer like the KDE Installer, and its not as friendly as Mandrake and others (dont pick at that please).
* This installer wont be a wrapper for redcarpet. End of story. Too much work and planning has already been done to just turn around and change ideas.
* If your not a newbie, or you dont like this idea, please dont talk about it (You know who I'm talking to). It seems no matter how many times I visit slashdot, at least 65% of your comments are put-downs.
I'm sorry if I seemed mad the last comment, my temper with flamers only goes so far.
Thanks, Nick Betcher
P.S. Topics like 'flamebait' are getting old. If its humor, its old :) Also, this article isnt written in anger.
Helix installer and KDE (Score:4)
My (pretty obvious) two cents.
A decent GUI front end to APT (Score:4)
* Make the code Open Source
* Make it cross platform, cross applciation (ie, and interface for apt), and cross licensing (this means closed source apps, too)
* Call it `Software Installer'. The command would be `installer'
* Allow cataloging (Eazel services, Red Hat network, and other subscriptions) into it. Allow people to make these services free (gratis) or otherwise.
* Make it secure. Put package signing into systems that don't have it yet.
* Have a set of packaging guidelines that go with it, defining things like granularity, package structure, menu icon guidelines, etc.
* If Oracle wants to, allow me to type in my credit card and download Oracle.
Re:Helix installer and KDE (Score:4)
signature smigmature
Re:This is a great move! (Score:5)
Sure it takes longer to build a program than it does to install it. But it ain't much harder (./configure; make; make install). And you get to be 100% in control. Your computer is yours. Freenix isn't Windows. Don't let the distros tell you what to do.
Checking for dependencies is a Good Thing(tm). Using it to install generic one-size-fits-all binaries optimized for the lowest common denominator is a Bad Thing(tm).
GRand Unified Installer?? (Score:5)
Frankly I'll never convince my sister to "./configure; make; make test; make install". It'll just never happen. Even "apt-get install" is too much for some people. Having a common GUI installer with a consistant look and feel would be REALLY beneficial to new users and wouldn't hurt us l33t h4x0rz either...
I'm of the general philosophy that more effort needs to go in to GENERIC tools and less into non-portable or app-specific ones, that way real, large-scale, code reuse can happen.
Justin Dubs
Oh NO!!! (Score:5)
I really feel bad posting this, GNU projects should be about community and sharing but you really should realise that this is one guy (that honestly has little clue) and is not the official KDE installer. My heart sunk when I saw this article, I do realise that KDE needs an installer but unfortualtly is a bit of a way off yet. For now, Debian apt is by far the best way to install KDE in binary form because of the way it's packaged but a lot of people don't run Debian. Compiling from source is obviously the most flexible route but is more trouble for those of us that want to get things done.
Again, this is unfortunate, please read the posts on dot.kde.org where the author has posted some very strange remarks but it backs this post up. I guess everything has a bright side though and this project highlights the easy of use of QT Designer.