Curtis Rey [Sat, 4 Oct 2003 05:42:31 -0500]:

./configure

make

checkinstall

Only if you are unwilling or unable to understand rpm spec files and the mechanisms involved in building software packages (automake, autoconf, libtool etc.) should you use a super cluge like checkinstall. Believe me, it's really rewarding if you can do without checkinstall as you have far more possibilities, including using a build root for installing a package so that you don't install into the running system when you build the package. Philipp -- Philipp Thomas work: pthomas AT suse DOT de SUSE LINUX AG private: philipp DOT thomas AT t-link DOT de

Carlos E. R. wrote:

The 03.10.05 at 01:30, Philipp Thomas wrote:

...

...

checkinstall

Only if you are unwilling or unable to understand rpm spec files and the mechanisms involved in building software packages (automake, autoconf, libtool etc.) should you use a super cluge like checkinstall.

I'm too lazy O:-)

...

Believe me, it's really rewarding if you can do without checkinstall as you have far more possibilities, including using a build root for installing a package so that you don't install into the running system when you build the package.

I believe you, of course! X-)

But unless I see a nice howto, of the "for dummies" type, I don't think I'll try.

I don't blame you. I've seen this format of thread repeated for as long as I've been using linux (about 5 years). Someone will as for a simple way to do something, and you will get replies that say, "All you have to do is..." followed by a complex task with not a single hint on how to do it besides rtfm. As often as this particular thread is repeated, I have yet to see someone give a 5 step approach as to how you can write a spec file and build an rpm in under 10 minutes, which is what I think you are looking for. Take it from me, you can safely use checkinstall as long as you remember that yast won't know how to upgrade YOUR rpm later on--namely because yast will depend on seeing the package using a specific naming format and so forth. But given how long it takes most distros to implement the latest version of anything, that probably won't happen anytime soon. Also, recognize that ensuring that all the dependencies are fulfilled is up to you. That shouldn't be a problem since tarballs always have a readme with the dependencies listed. If you feel comfortable with that, go ahead and use checkinstall. It is by far the easiest way to make an rpm. It also makes uninstalling a snap. That way, if you ever want to switch to a suse package later, you just uninstall your own with rpm first. Unfortunately, many linux people take the time to learn something then assume that everyone else has that much time on their hands. After it becomes easy to them, they expect everyone else to find it as easy as they do. Not that learning to write a specfile would not be worth your time, but perhaps you feel your time could be better spent. If this is the case, you can safely ignore any comments to the contrary. Just my 2 cents. John S.

On Mon, 06 Oct 2003 00:33:59 +0200 Philipp Thomas wrote:

Even I need at least between half an hour and an hour to get a spec file right.

Even you, Philipp? My God, what hope for the rest of us?

Somehow people always think that complex tasks should be made easy

Why not? That's the way progress is made. Did you produce the plastic and everything else to make your computer, your monitor or printer? Or did you do it the easy way?

Building rpms is a skill you have to learn, but which will pay off handsomely.

No, we don't "have" to learn, we may cloose to do so, and, yes, it may pay off. But don't bet on it being the case that the results outweigh the effort. Terence

In a previous message, Philipp Thomas wrote:

Any effort that gets rid of checkinstall is IMO worth it.

I know that you're half joking with this, but all this ranting against checkinstall is stupid IMO. This program has a specific and very useful purpose - to enable the quick and dirty insertion of a compiled program into the rpm database. For an individual user, there are no real disadvantages to its use for building software, and it makes it far easier to remove the software or upgrade it when the time comes. Sure, it's not as nice and elegant as actually getting a spec file correct, but it's also *far* less effort. If you are (for example) only compiling the app to see whether you like it, do you really want to spend half an hour (or more) of actual human effort? It's horses for courses. 2p John -- John Pettigrew Headstrong Games john@headstrong-games.co.uk Fun : Strategy : Price http://www.headstrong-games.co.uk/ Board games that won't break the bank Valley of the Kings: ransack an ancient Egyptian tomb but beware of mummies!

Philipp Thomas wrote:

js [Sun, 05 Oct 2003 20:02:27 +0200]:

...

As often as this particular thread is repeated, I have yet to see someone give a 5 step approach as to how you can write a spec file and build an rpm in under 10 minutes, which is what I think you are looking for.

Sorry, but that's impossible if you want to do it right. Even I need at least between half an hour and an hour to get a spec file right.

Right. Now take the average user and you can at least triple that. If he's a recent convert from windows, far longer if at all.

...

But given how long it takes most distros to implement the latest version of anything, that probably won't happen anytime soon.

Oh, normally we do with every new distro, if there aren't reasons not to upgrade.

Some packages, yes. I remember getting my new copy of suse 8.0. I installed thinking I would have new samba features, but the versions shipped was several months old and many release points behind. I had to go to samba.org just hours after installing a suse I had pre-ordered. This is not a complaint mind you. I know you want to test critical packages and ensure you ship stable versions, but it only explains why someone might want to move a little quicker and compile a new one themselves.

...

Unfortunately, many linux people take the time to learn something then assume that everyone else has that much time on their hands. After it becomes easy to them, they expect everyone else to find it as easy as they do.

No, I don't expect that. But you don't learn to drive a car in one day either. Somehow people always think that complex tasks should be made easy ....

Uh...isn't this what computers are for??? I shutter to think I spent 1500 bucks to make things more complicated.

Building rpms is a skill you have to learn, but which will pay off handsomely.

Philipp

You are no doubt correct on this point. I will not dispute it. I only point out that there are always exceptions, and you have to admit you make it sound far easier than it really is. When I explain something to people I always ask myself if I would have understood it 1 month after installing my first distro. If not, and I don't know for a fact I'm dealing with an advanced user, I try to make those complex tasks as simple as possible while suggesting that he/she look into the tougher method whenever possible (if waranted). I absolutely never discourage using the easy way because "No pain, no gain" is a great motto in the gym, but it sucks when you're trying to get your work done. John S.

Ivan Sergio Borgonovo wrote:

On Mon, 06 Oct 2003 02:08:11 +0200 js wrote:

...

<snip>"No pain, no

...

gain" is a great motto in the gym, but it sucks when you're trying to get your work done.

"No pain no gain" is perfect. It is just the coefficient that may be different.

K*pain=gain

Some have enough experience (past gain) or are smarter enough to have large coefficient in some tasks so that the pain pass unnoticed.

Coefficients??? Are we talking about computers or calculus? Again, if the pain or the gain gets in the way of doing your business, it's an obstacle. Any IT manager considering a switch to linux will tell you that learning curve is a very significant factor because it means lost productivity. Forget "No pain-no gain" and try this motto: "Work smarter, not harder."

I wouldn't say as simple as possible but as simple as convenient. Would you make a F1 Ferrari (or McLaren or Williams just to be a bit more unbiased) as easy to drive as a Renault Twingo?

From coefficients to Ferraris? Sheesh! What was the subject again?

Pretending Linux is simple for every situation (as saying it has 0 TCO) is a major mistake in Linux advocacy.

I don't care about advocacy, just getting the job done. Read any interview with Linus and he will say it himself over and over. Check out what linux says about this in an interview on kde.org. Note the reference about users who don't want to know what's going on under the hood. He says they are just as big a help to linux. Read: http://www.kde.org/history/linus.php "HY: We have seen many distributions of Linux that allows users to install Linux without knowing what's under the hood. While this has brought in tremendous new users to Linux, there are people who claim that this undermines the spirit of freeware because people are never forced to look under the hood and understand its workings. Is this a concern for you? Linus: No, I think this is only for the best. I don't think everybody should be interested in how an operating system works: it happens to be what _I_ am interested in, but I also think that any program is only as good as it is useful. So a useless program cannot be good, regardless of _how_ well it is implemented. The fact that there are lots of Linux users who don't care how the kernel works but only want to use it is not only a tribute to how good Linux is, but it also brings up issues that I would never have thought of otherwise. Those users tend to do different things from what I do, so their needs are different. And in many cases those differences have shown something that was missing or badly done in Linux. So even though these users aren't interested in how Linux works, they have been instrumental in making it better. " So. I guess if you will argue with Linus, there is NO WAY you will EVER listen to me. So I'll leave it at that. John S.

js [Mon, 06 Oct 2003 02:08:11 +0200]:

but it only explains why someone might want to move a little quicker and compile a new one themselves.

I do understand that point, having done so a lot before I joined SUSE ( I just don't have that time anymore). And had a tool like checkinstall existed when I started using linux back in 95, I guess I would have used it.

...

Somehow people always think that complex tasks should be made easy ....

Uh...isn't this what computers are for???

To a point, yes. But I have a number of things in mind where the computer causes more problems than it solves. And as I wrote in an earlier mail, what I forgot to add that there is a limit as to how easy it should be made.

I absolutely never discourage using the easy way because "No pain, no gain" is a great motto in the gym, but it sucks when you're trying to get your work done.

I doubt that you really *need* the newer version of a given package to get your work done. I guess many of those wanting to update are version junkies like I myself once was. In the years at SUSE, where I work with Linux all day, I have had only a very few cases where I definitely needed a newer version of a given package because of additional features the newer package offered. Philipp

Believe me, it's really rewarding if you can do without checkinstall as you have far more possibilities, including using a build root for installing a package so that you don't install into the running system when you build the package.

Not to mention being able to split a package into the main package, development headers, plugins (if appropriate) etc etc... -- James Ogley, Webmaster, Rubber Turnip james@rubberturnip.org.uk http://www.rubberturnip.org.uk Jabber: riggwelter@myjabber.net Using Free Software since 1994, running GNU/Linux (SuSE 8.2). GNOME updates for SuSE: http://www.usr-local-bin.org

Now that you mention it: I also left out %prein/%postin and %preun/%postun, i.e. the possibility to do additional things the software package itself does not do:

... Provides: Obsoletes: Requires: Oh man, the list of why not to use checkinstall just goes on and one don't it? :) -- James Ogley, Webmaster, Rubber Turnip james@rubberturnip.org.uk http://www.rubberturnip.org.uk Jabber: riggwelter@myjabber.net Using Free Software since 1994, running GNU/Linux (SuSE 8.2). GNOME updates for SuSE: http://www.usr-local-bin.org

...

./configure

make

checkinstall

Only if you are unwilling or unable to understand rpm spec files and the mechanisms involved in building software packages (automake, autoconf, libtool etc.) should you use a super cluge like checkinstall.

Believe me, it's really rewarding if you can do without checkinstall as you have far more possibilities, including using a build root for installing a package so that you don't install into the running system when you build the package.

Philipp

I think he wanted to make an rpm. In less than six hours.

Philipp Thomas wrote:

And? It's not that hard to learn the basics of package building, otherwise there wouldn't be sites like pacman and lots of volunteers building rpms for various distributions.

If, for instance, the package has an init script, you will have to modify it anyways, as 98% are written for redhat and thus won't work correctly (if at all).

checkinstall's main feature is giving you a false sense of security.

Philipp

No, it is quite hard to learn the basics of package building otherwise everyone would be at it. I really don't think that someone with, say, one week's experience of Linux will be merrily writing their own spec files and feelding in patches like an old hand. It would be a lot easier to build suse-specific packages if suse released a packaging howto that was written with less experienced users in mind. There is some material already, but it's written with highly experienced users in mind. Until then, I guess my overall impression is that suse quite like the idea that it's no cakewalk to build packages for their distribution and that it's better to stick with what Auntie Suse has provided for you. Possibly, there is a hard-nosed business calculation behind this too, i.e. securing orders for upgrade editions. Perhaps the upshot is that a lot of folks will happily continue to use checkinstall and if the stuff fails to work, well, yes, some may perhaps hold it against suse. :) Mark

The more fool them then, particularly given the lengthy and patient explanations from SuSE staff as to why it's not a good plan.

and non-SuSE staff (well, ex-SuSE staff in my case) My point being, it's not a point that's being made for nefarious purposes to extort money from users, these are valid technical arguments that myself, Philipp and the others have given for doing it properly, rather than with checkinstall. If someone is capable, and confident to build applications from source, then it's really not a huge step to learn to craft a spec file too. Hell, if it's an update to a package that's already in SuSE, base it on the spec in the SuSE src.rpm, if not, pick another package that seems similar (by whatever means of comparison you like - ie, are they both library packages, or are they both GNOME packages, etc, etc...), and base it on that. If checkinstall was a good way to build RPMs, the distro you buy would be built with it. It isn't. -- James Ogley, Webmaster, Rubber Turnip james@rubberturnip.org.uk http://www.rubberturnip.org.uk Jabber: riggwelter@myjabber.net Using Free Software since 1994, running GNU/Linux (SuSE 8.2) GNOME updates for SuSE: http://www.usr-local-bin.org

Mark [Mon, 06 Oct 2003 09:30:20 +0100]:

I really don't think that someone with, say, one week's experience of Linux will be merrily writing their own spec files and feelding in patches like an old hand.

I don't say that either. But those people should IMO just stay away from trying to install their own packages, as otherwise they risk messing up their system royally without knowing that they do.

It would be a lot easier to build suse-specific packages if suse released a packaging howto that was written with less experienced users in mind.

There is a limit as to how easy you can make it. For instance you need a knowledge of the basic development tools to really understand why certain things are necessary or desirable. There is no simple recipe that will fit all packages. And who is going to pay for that?

Possibly, there is a hard-nosed business calculation behind this too, i.e. securing orders for upgrade editions.

Nonsense! It's simply not that easy and requires a certain amount of knowledge, something you can't provide with a short HOWTO.

Perhaps the upshot is that a lot of folks will happily continue to use checkinstall and if the stuff fails to work, well, yes, some may perhaps hold it against suse.

If somebody willingly breaks his or her system, they get to keep the pieces and complaints will be rejected, as we offer no support for building your own software (well, we do, but at EUR 1600 per man day). Philipp

On Mon, 06 Oct 2003 13:26:52 +0100 Mark wrote:

Nice work if you can get it! Hehe. I think all you guys do a fantastically good job but maybe you're getting caught in a no-win as Linux user patterns change. Judging from news groups, a lot of new users head straight for the multimedia section and then start asking how to build or rebuild the apps to provide all the ripping, playing, encoding, etc. stuff that they've come to take for granted on er other

What is granted? Is it granted you can build up Windows Media Player optimized for your architecture? Have you ever read the EULA? Have you ever read the EULA of most shareware/addware/spyware expecially filesharing proggies and media players? Come on... such stuff works out of the box in Linux too, modulus some license for MP3 and such kind of stuff. If people what to be bleeding edge they have to pay for it, in terms of time if they have or in terms of money if they have. I hope I'll never see SuSE Personal, SuSE Professional, SuSE Dumb. Consider every upgrade of Office cost you a lot of Euro (and a bit more USD <g>)... keeping yourself on the bleeding edge for commercial software is quite expensive... and Windows shareware/freeware is not that bleeding edge. Beside the fact that as a customer of SuSE I like to put pressure on them to give me what I like, I've to make reasonable requests. I would like they will put pressure on other vendors to make Linux and RPM distributions more and more standard (cos I don't like to be obliged to be faithfull). I would like they definitively change YaST license. I would like they reduce the number of "who know them" packets and keep more uptodate eg. OpenOffice. What I like is they invest in KDE, kernel (hi Andrea), gcc developement. I appreciated they added more packages to the download section so now you can keep uptodate KDE, X, Samba, Apache and Gnome... Can we ask some more? ;) just 3 or 4 more ;) At 2h for package at 1600 Eur/day it shouldn't be a too huge investment for SuSE ;) You'd put up a voting system for that. My youth is gone, the period when I enjoyed to try everything passing nearby my PC died afer a long agony... Having bleeding edge rpm is not anymore a priority, response time for security patches IS. I bet that if SuSE is really interested in further standarization of Linux, building up your own rpm will get easier and easier... OK back to something less phylosopical

On Sun, Oct 05, 2003 at 02:08:31PM +0200, Ivan Sergio Borgonovo wrote:

Believe me, we don't want to steal your job but is there a SuSE HOWTO to help people build their own S/RPM specially tuned for SuSE? or just a list of things you've to take care of when building RPM for SuSE?

http://www.suse.de/~mmj/Package-Conventions/ Regards, -Kastus

Op zondag 5 oktober 2003 20:33, schreef Ivan Sergio Borgonovo:

Is there an HTML index/searchable index to the content of ftp://ftp.suse.com/pub/people/ ?

Most of SuSE work and SuSE employees work should be much more advertised.

To start with there is this page: http://linux01.gwdg.de/apt4rpm/freshrpms.html Search for suse-people and you have an idea what has been changed in the last 10 days. Further more there are contents files that show for each suse version since 7.3 which packages are available. These are just flat data files, and can be easily turned into html pages. Maybe better have these changed into searchable web pages. The data is there, the frontend is missing. Perhaps a happy suser (suse user) can pick this up? How to use the contents files can be used is explained here: http://linux01.gwdg.de/apt4rpm/home.html#contents -- Richard Bos Without a home the journey is endless

Op zondag 5 oktober 2003 21:39, schreef Ivan Sergio Borgonovo:

Well for me is more than enough. It is just a pity I've to download a so large file just to see if the few packets I like to keep uptodated have been updated... not for my bandwidth... but for theirs.

OK... not terrible... the gzipped list is 45Kb for 8.1.

Use rsync in that case, is should save bandwith. If want bigger files providing information about rpms download the AptContents files from this directory: ftp://ftp.gwdg.de/pub/linux/suse/apt/ They provided information about which file is being provided by which rpm. Very handy if things don't want to run on your system. As said before it would be nice to have a html frontend to this file as well. -- Richard Bos Without a home the journey is endless