Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Linux Software

What Package Management Features Do You Value? 70

0x0d0a asks: "Slashdot has now had a number of articles on package management. Strong opinions about the management approaches of Red Hat, Debian, Gentoo, Slackware, and BSD have all been expressed, some quite negative. What suggestions do you have for improvement? What features do you value in a package management system, and in what areas would you like to see additional functionality?"
This discussion has been archived. No new comments can be posted.

What Package Management Features Do You Value?

Comments Filter:
  • easy. (Score:4, Insightful)

    by gl4ss ( 559668 ) on Friday January 03, 2003 @06:03PM (#5009794) Homepage Journal
    it has to be easy in overall.

    easy to install stuff the first time(type one line or press one button and it'll figure out the rest).

    easy to remove that stuff without it leaving other stuff unworking.

    easy to keep up-to-date.

    well.. apt-get fits this bill at the moment for me.. i don't care much of compile-locally-optimized-whizmo jizmos.. nor don't i think that downloading binaries from debian is a security concern anymore than downloading sources through some portage system(heck, i'm wouldnt check the source anyways).

    and i find dselect comfortable to use and easy to find software from..
    • Everything you said is awesome, but. I think being graphical is an absolute must.

      My ideal package manager would be where everything installs graphically, like in windows, or like mozilla. It installs in a location in the filesystem specified by the user. And this should be predictable, no making extra folders in folders like windows apps do. If something has dependencies it should check for them and automatically upgrade/install them if they are not present without ruining compatability for any other app.

      There should also be a seperate piece of software with a list of all installed packages organized by type of application. The packages should have real names like "Mozilla Version 1.2.1" not "mozilla-i686-pc-linux-gnu-1.2.1-sea". From this graphical application I should be able to automatically upgrade install and remove new packages. The program should also "just know" where to get updates to packages and how to install them.

      Not only would this be a vast improvement in ease of use over current linux options, but it's easier than windows. And the only thing I've seen yet that's easier than windows is knoppix. So it's saying a lot.
      • Re:easy. (Score:3, Insightful)

        by toast0 ( 63707 )
        how are you supposed to install the gui with a graphical package manager?

        why not have the package manager be all unixy and easy to abuse with pipes and such, so that if somebody wanted a graphical front end, it could happen, but if somebody wants to just maintain their server through a terminal, they could too?

        • Provide just enough extra code in the graphical package manager to run in 640x480x16 VGA mode.
          • 640x480x16 vga mode doesn't work so well if you don't have a vga card :)

            (yes i'm being a prick, but some people legitimately use serial consoles for things, and a graphical package manager wouldn't work well for that)
  • It has to detect.. (Score:4, Insightful)

    by t3kad0n ( 636763 ) on Friday January 03, 2003 @06:08PM (#5009841)
    It has to detect if the libraries it needs have the same functionality as the newer ones I have. Needing version .40 of a library and not accepting library version .50 if it works the same as version .40. My perfect package manager wouldn't take many hours of frustration to make your own packages. :)
  • Debian Policy! (Score:5, Insightful)

    by reynaert ( 264437 ) on Friday January 03, 2003 @06:08PM (#5009843)
    The Debian Policy manual [debian.org] is the reason behind apt-get's magic.
    • The policy manual is one of the main reasons I use debian. I can look up exactly where something should go and what it's supposed to do instead of digging around trying to figure out what some random developer thought was the 'right' way to do it. If only the LSB were as good as the policy manual, then we wouldn't have a problem with random distros.


  • Something as important as a package manager can't be this buggy [redhat.com].

    • The rpm version that ships with RH 8.0 is pretty buggy and can lead to lockups. RH has an unsupported fixed version of rpm [redhat.com] that fixed these problems for me (I was only impacted by one of the bugs). They haven't put out an official update yet.

      Note: do *not* upgrade to the version of RPM in Rawhide or Phoebe, RH's current beta. It's a *pain* to get back -- you get moved to hash-version 8 if you --rebuilddb at any point, and db4 as packaged by RH doesn't grok hash-version 8. You'll need to grab db4 from Sleepycat and use it do dump your rpmdb and reload it with RH's db4 if you want to get back (this happened to me).
  • by Froze ( 398171 ) on Friday January 03, 2003 @06:12PM (#5009891)
    Except that there needs to be a catagory entry. What I mean is a way of getting all similar types of packages. For instance suppose I wanted to look at all thing "word processing", then I would get packages ranked from most applicable (open office, abiword, etc) down to quasi applicable ( vi, gnotepad, hexedit, etc).
  • What I've Loved (Score:5, Interesting)

    by MBCook ( 132727 ) <foobarsoft@foobarsoft.com> on Friday January 03, 2003 @06:15PM (#5009923) Homepage
    I've used many many distros over the last few years, and I can tell you the things I like easily. Below are some of my top ones:

    • Dependencies - This is probably the most important for me. The thing that makes apt so great is that it can do dependencies. Gentoo's emerge also does a great job of this. I haven't used an RPM distro in about 2 years, but back then they didn't do anything but complain that you needed some other package. YOU had to go find it. YOU had to go install it. YOU had to get IT'S dependencies, etc. It meant installing one package could take forever.
    • Source - I like being able to easily build from source. With RPMs (at least in my expirence) it would build the package, then put it in some odd location and you'd then have to "rpm -Uvh" (or whatever) from there. Gentoo does a great job of this, but it's a source distro ;). Basically, when I install from source, it should install the package for me. If I only want to make some kind of binary that I can distribute, that should be a seperate command.
    • Compatibility - RPMs never seemd to work across distros, quel suprise. This is one thing that I really like about slackware's .tgz files. They are nothing but a .tar.gz with some extra info, so no matter what system you use, you could just download the slackware .tgz and use it, right? Gentoo doesn't have packages, but "ebuilds". These are nice because they are small little text files, and your computer goes and fetches the latest version of the package (or whatever version is specified). It uses the standard source and it gets it the same way you might.
    • The Unistall - This can be a PAIN. This is the one feature that, IMHO, makes packages better than source. If this doesn't work, why not take the extra 3 steps to use source? When I uninstall something it should be removed completely. No empty directories, it should offer to remove it's config files or back them up, it should offer to restore any files that it's changed, etc. Both Debian and Gentoo do a great job with this. I don't remember what it's like with RPMs very well.

    I'll post more if I can think of them. Why does constructive criticism have to be so much harder than normal criticism? He he he. I talk alot about Debian and Gentoo because those are the two distros that I use regularly, and the package systems are a big reason for that. Packaging makes a difference. I'd probably run Mandrake if it wasn't RPM based. It's a great distro, but I just CAN'T STAND RPMs. Are they much better now than a year or two or three ago? Almost certanly. But I've been so soured to them by my expirence, it will be quite a while before I try them again; especially since I found apt and emerge.

    • # Compatibility - RPMs never seemd to work across distros, quel suprise. This is one thing that I really like about slackware's .tgz files. They are nothing but a .tar.gz with some extra info, so no matter what system you use, you could just download the slackware .tgz and use it, right? Gentoo doesn't have packages, but "ebuilds". These are nice because they are small little text files, and your computer goes and fetches the latest version of the package (or whatever version is specified). It uses the standard source and it gets it the same way you might.

      I can understand and even live with binary RPMs not working across distros. I find it really annoying, however, that *source* RPMs are frequently not buildable on other distros. Mandrake's and PLD's frequently don't rebuild for use on RH, which is quite frusterating. They use slightly different macros in their .spec files...

      This is an area where standardization would make things *much* better, particularly since the differences between the formats isn't that big.
    • Tip for RPM users (Score:5, Informative)

      by 0x0d0a ( 568518 ) on Friday January 03, 2003 @06:47PM (#5010223) Journal
      RPMs never seemd to work across distros, quel suprise. This is one thing that I really like about slackware's .tgz files.

      And some things aren't RPM-packaged.

      One tool that *no* RPM user should be without, IMHO, is checkinstall [asic-linux.com.mx]. This runs a normal "make install" after you're done with ./configure and make, but monitors what locations files are being "install"ed to. It then builds an RPM package and installs it. This lets you cleanly uninstall tarballs, and handles library dependencies. In many ways, it gives you the flexibility of Slackware's approach with the nice features of RPM.
  • by cornice ( 9801 ) on Friday January 03, 2003 @06:16PM (#5009938)
    I've had success with RPMs, DEBs, EBuilds, etc. What really makes the biggest difference is the packager. Most major distros have pretty good package maintainers now. This wasn't always the case.

    Now for me it's all about convenience. If I can use Debian, Gentoo or Mandrake+Ximian Red Carpet and keep my system up to date on a daily basis then I'm happy. This requires good packages and good mirrors. I throw Ximian in there becuase I've had a terrible time with Mandrake mirrors. Also if I can upgrade without running an install from CD then I'm happy. Debian and Gentoo seem to do this quite well. If I can avoid conflicts and install a couple versions of the same thing and keep it all straight, then I'm really happy. Gentoo seems to be making strides towards the last one but compiling everything isn't always an option.
  • I value a package manager that uses a decent naming convention for its packages. If I want to install ncurses, I want to type install ncurses, not install libncurses5.

    Also, I like a package manager that does a decent job removing packages. If I want to completely be rid of pptpd, all I should have to type is remove pptpd, and not have to worry that it left a million directories and configuration files sitting around.

    Finally, I prefer a package manager that lets me override dependencies if I wish. For example, say I want to use pptpd. This requires ppp, but the version of ppp installed is only 2.0. I need 2.1, so I'd like the ability to somehow force the package manager to ignore the fact that ppp isn't installed so that I can install pptpd through it without having two versions of ppp cluttering things up.

  • by Tuxinatorium ( 463682 ) on Friday January 03, 2003 @06:29PM (#5010070) Homepage
    Well, I prefer Fedex for my packages... UPS sucks and priority mail is too slow.
  • by tubabeat ( 605286 ) on Friday January 03, 2003 @06:30PM (#5010081)
    There's little point standardising on rpm (because thats what the LSB says) or any other package management system unless packages are named consistently. I can't pick a Red Hat rpm and have confidence that I would be able to install it on Mandrake or SuSe or ... because I can't be sure that the dependencies will be satisfied as some packages have different names (one example perl-base-5.8.0 on Mandrake is called perl-base-5.800 on RedHat, enough to screw up a dependency check)

    This doesn't really bother me because I'm quite happy to build from source (or even make my own rpms if I've got many installs to do) - but to be honest unless a package management system can be relied upon to work always then its isn't working properly and that is a bug (even if its a design bug rather than an implementation one) not a feature.

    I guess maybe a solution is for rpm (or whatever) to save version numbers for each file it installs and for dependency checks to be strictly only on files rather than packages. A more sophisticated rpm wrapper (like urpmi) could then map failed dependencies back onto the appropriate package for the distribution. I suspect this is nowhere near as trivial as it sounds!
    • There's little point standardising on rpm (because thats what the LSB says) or any other package management system unless packages are named consistently.
      I'd say that is a specific problem of RPM (or the way it's used in practice). It could simple look for an executable perl$PERLVERSION in $PATH, or even test the output of perl --version or something.

      On the other hand, you'd better know when to stop with this, or you'll end up reinventing autoconf...

      • I'd say that is a specific problem of RPM (or the way it's used in practice).

        You're right, it is a specific problem with the way its used in practice. But on the other hand it could also be argued that the fact such a situation can occur is down to flawed design.

        It could simple look for an executable perl$PERLVERSION in $PATH, or even test the output of perl --version or something.

        It could but that would be horrendous to implement so that it worked for everything any package might depend on (-v --version -version etc. etc. and then parsing the output!).

        I'd be interested to know whether this is just a problem for rpm (after all rpm is probably supported by more distros than any other solution)or whether other package management systems suffer the same problem. For example can you install Debian packages on other distros that use apt-get (without running into dependency problems etc.)? P.S. I quite like rpm (but then I have used Solaris pkgadd, at least with rpm the package names give some clue about what they contain!)
        • It could but that would be horrendous to implement so that it worked for everything any package might depend on (-v --version -version etc. etc. and then parsing the output!).

          It could revert to the "check if this RPM is installed" if any other test is too difficult.

          Checking directly to see if a new enough version of a shared library exists, rather than checking to see if it's RPM is installed, would be a huge win.

  • uPM... (Score:2, Informative)

    by DarkBlack ( 5773 )
    Many may not have had the pleasure of using uPM [u-os.org] yet. It is a new package manager that seems to be very well thought out and seems to be heavily bassed on dpkg [debian.org] and ebuild/emerge [gentoo.org]. It represents the best features of both. It aims to simplify the build process and allow for binary packages to be built from the customized build process.

    I would look for uOS [u-os.org], the distribution based on the package manager to rise above either Debian or Gentoo (both of which I love, btw). It is still early in development, but it seems to work well for me.
    • I took a look at this a while ago -- the features list sounded really, really nice, but I'm waiting for something more mainstream to adopt it first.
  • I guess it is rather important that actually creating the packages is easy, so that I as a mere user am likely to not have to bypass the package management system when I happen to want to use some oddball software, or a newer version than my distro/OS includes.

    That is one of the reasons for me to like *BSD so much: Writing a port is not much harder than saying where you got the sources from and what files it does install (and that part can be automated, even), so it actually saves work keeping a nice, clean system. It may be just me, but I never bothered to build my own RPMs or DEBs when I was using Linux.

    • checkinstall (Score:4, Informative)

      by swillden ( 191260 ) <shawn-ds@willden.org> on Friday January 03, 2003 @07:04PM (#5010358) Journal

      It may be just me, but I never bothered to build my own RPMs or DEBs when I was using Linux.

      I nearly *always* build DEBs for my Debian boxes (well, for the occasional app that Debian hasn't already packaged), and I've never bothered to learn how debs are made. How? checkinstall

      To use it, you just run:

      ./configure <appropriate opts> && make && sudo checkinstall

      checkinstall will run "make install" for you, but will do it in a chroot environment, see what got installed where, build you a DEB that will do it and then run "dpkg -i" to automatically install the DEB for you.

      And, of course, "aptitude purge <pkgname>" will get rid of it all.

      • you can also run dpkg --purge [pkgname] which will save aptitude the work of invoking dpkg :)
      • The parent post has to be the most useful post i've seen in months. Thanks Swillden!

        I immediately started searching for an RPM equivilant and found that checkinstall works for RPMs, .debs, and Slackware.

        FYI, checkinstall can be found here [freshmeat.net]. This and apt4rpm [sourceforge.net] makes admining Redhat machines a lot more fun.

        • Thats fantastic, I wish I'd seen that before. If this is as good as it looks (I wonder how it gets on with dependencies etc.) then I'll be rpm'ing and archiving every source install I do - great if I need to do a fast rebuild of something.

          Mandrake packages are on rpmfind, you need both checkinstall [rpmfind.net] and libcheckinstall1 [rpmfind.net] [Note they seem to have a circular dependency so you need to install one --nodeps]
          • I wonder how it gets on with dependencies etc.

            It ignores them.

            Note they seem to have a circular dependency so you need to install one --nodeps

            Been a long time since I used RPM, but... I believe you can install both of them with one invocation of rpm -i, and then you don't have to do the nodeps thing.

          • (I wonder how it gets on with dependencies etc.)

            It creates dependencies at install time for stuff that can be determined with ldd...so if you try to uninstall a library package that the new package depends on, RPM will complain.

            Granted, this doesn't help if the packaged program calls an external program through fork()/exec(), but it sure beats nothing.

            Mandrake packages are on rpmfind, you need both checkinstall [rpmfind.net] and libcheckinstall1 [rpmfind.net] [Note they seem to have a circular dependency so you need to install one --nodeps]

            Weird -- can't see how libcheckinstall would depend on checkinstall. Anyway, you can handle circular dependencies by installing both at the same time. RPM will work out whether things are kosher.
        • For reference, not for one-up-manship, Gentoo has this built directly into the emerge/ebuild system.

          A side effect is with only a little bravery, you can usually slightly modify an ebuild to work with a new version of something, sometimes as easily as just renaming the ebuild file, and still be very confident that the uninstall of the old version and install of the new version will be perfectly correct, despite the fact that no "official" distro guardian is even aware of your ebuild.

          Many people are not aware these things exist, but given their wide (if quiet) availability, users should consider the use of some kind of sandbox during installation a requirement for any distro management system they are considering.
  • Better front ends (Score:3, Interesting)

    by 0x0d0a ( 568518 ) on Friday January 03, 2003 @06:41PM (#5010167) Journal
    I'd like to see front ends be a bit better. I was quite impressed with apt-get, and use it with RPMs on a RH system. RH's up2date is really awful -- it's sluggish, doesn't give much feedback on what's going on, fragile (rpm hangs or a download fails, and it isn't very smart about it), doesn't resume failed transfers, and doesn't let you download non-RH packages. It simply feels flaky, which is not good for a tool that, for many users, may be their only look into the management of their system.

    Oh, and it grabs an exclusive lock on the rpmdb the entire time the thing is downloading. I *really* think this is a bad idea -- at the very least, there should be an option to flip this off. Novice users aren't going to be running rpm manually anyway at the same time, and more experienced users *really* get annoyed when they can't query or modify in unrelated ways the system while up2date is slowly sucking something down over a modem.

    Apt-get is nice...if there isn't something like it, it might be a good idea to have a Red Carpet-like GUI for it to make it really appealing to new users. Hell, most people don't use Windows Update because they consider it too intimidating or don't know about it...
    • i think apt-get, aptitude, dselect, and dpkg all (attempt to )grab the lock on the debian package database when they're running, regardless of if they're actually doing package stuff, waiting for user input, or downloading.

      i'm sure there are flags to dpkg and rpm to ignore the lock if you really wanna do two things at once
    • ...try using the "synaptic" front end from connectiva web page or google around for some other places. Great apt4rpm front end gui tool.
    • by ctr2sprt ( 574731 )
      As someone else noted, all of Debian's tools also do the same dumb blind-locking. Extremely annoying. That needs to go out the window: lock when you need it, never at any other time.

      Next up, apt-get is bad about handling low disk space. Try apt-get upgrade when you're going from stable to unstable. You need to download 100MB+ of packages for a reasonably complete install. That's more than many people have in /var, which is where apt-get stubbornly insists downloaded files must go. If there's a way to change this, it's undocumented, because believe me I've looked. So apt-get needs to be smarter about downloads. There's not 100MB available? Fine, figure out how much is available and download that much. Install it, delete it, and then download the next chunk. Low disk space situations can actually cause serious problems because dpkg apparently doesn't check free space correctly when it's installing packages. It unpacks it into a temporary directory in /var, runs the configure script, and then tries to copy it to the final location. This is the right way to do it, except that because there's no disk space checking, you can run out of disk space anywhere during that process and get in trouble (like if it happens when replacing libc, for example). So package managers must check to make sure there's enough free space every step of the way, and it must be able to roll back the part of the install that it had already done. I don't expect this to be totally perfect - there will always be race conditions on a multiuser system here - but anything is better than what dpkg has now.

      Another important thing is smarter handling of version numbers in the package database. Debian tends to suffer from problems related to this, which is why you see packages named "lib2" and "lib3" (e.g., two completely distinct packages, rather than two packages which happen to have different versions). The reason for the double packaging is that often a program relies on a specific version of the library and it's convenient to have both libraries installed at once. But the package database and versioning system doesn't support two identical packages with different numbers: the software just blindly tries to install the newest one, and bails if it can't do that and still resolve all dependencies. This causes all the sorts of problems RPM users often see, which is two functionally-identical but differently-named packages causing dependency errors, loops, or other problems. It also draws out the single most frustrating aspect of package management for the user, which is that they don't understand why they need a package and therefore don't understand the relevant differences between, say, glibc-2.3.1 and glibc-2.2.4. They usually don't even realize that those are the same package, just two different versions, and so they try to install both... usually with --force flags and disastrous results. Package managers, and packagers (the people), need to get rid of the ambiguity and make sure that users can always say "glibc" and let the package manager worry about the version numbers. This is what Debian was like for a long time, and I'd like to see it be that way again.

      "Meaningful defaults" is also a good thing here. I like this about FreeBSD and RedHat. The configuration scripts aren't shy about making decisions for you. That's fine with me, as long as those defaults are sane (usually they are). What I dislike is when the package maintainer starts getting involved in policy (e.g. Debian and OpenSSH), or when the configure script wants its hand held every step of the way (default setting in Debian). Packages should also never have defaults which go against the standard system policy (e.g. Debian's OpenSSH enabling root logins by default, even though FTP, Telnet, and all other similar services do the reverse by default). There also needs to be consistency in the defaults for a package between versions, so users don't get caught off-guard by the change (lots of packages, sadly, are guilty of flip-flopping between defaults when they aren't sure which to use). If all else fails, put a syntax error in the configuration file to force the user to edit it by hand.

      This may sound more like criticism than positive things, but I think these emphasize the good parts of package managers. Most of my complaints are about features incompletely or poorly implemented. I basically like working with packages on any modern Unix(-like) system. It's pretty trouble-free and usually Just Works. Most of my complaints are nitpicks or things that bug only power users; in other words, things to improve on in the future, since the great bulk of features have already been implemented and are working great. I think the reason for all the criticism, in my post and others', is that there're so many good aspects of modern package managers that it's shorter to list the defects than the great, useful, working features.

      • "Next up, apt-get is bad about handling low disk space. Try apt-get upgrade when you're going from stable to unstable. You need to download 100MB+ of packages for a reasonably complete install. That's more than many people have in /var, which is where apt-get stubbornly insists downloaded files must go. If there's a way to change this, it's undocumented, because believe me I've looked."

        I have also run into this situation (think firewall box with tiny old 400mb hd). I just mounted a network share, pointed a simlink (/var/cache/apt/archives) to it and went on my happy way. Could be more difficult if the box is not networked, but those seem to be quite rare in Linux land. If you have the disk space in another partition; even easier!

        As to locking, agreed! That is very frustrating!

      • Next up, apt-get is bad about handling low disk space. Try apt-get upgrade when you're going from stable to unstable. You need to download 100MB+ of packages for a reasonably complete install. That's more than many people have in /var, which is where apt-get stubbornly insists downloaded files must go. If there's a way to change this, it's undocumented, because believe me I've looked.

        I believe you, but it's not undocumented. Try apt-get -o dir::cache::archives=/some/other/dir ... or setting the Dir::Cache::archives setting in /etc/apt/apt.conf

        Or, like the other poster said, symlink the archives directory to the other partition. I don't have an easy answer to the other parts about apt-get's desire to download and upgrade all at once, but others might.

        • I've seen other people talk about those strange apt-get configuration settings. Where are these all documented? There's passing mention of a handful of them in the apt-get man page, but no reference to where the complete list is. Actually, I just figured that was the complete list.
    • I'd like to see front ends be a bit better.

      I do not understand this, but graphical package managers have been the one single consistent failure for me in Linux, since Redhat 4.2.

      Every single graphical package manager I've ever tried, with the emphasis on tried, to use has failed miserably for one reason or another. Red Hat 4.2's package manager seg faulted. In my Debian era, the Debian graphical managers were the only programs to consistently crash on me, once even leaving the system databases in an inconsistent state. kportage on my current Gentoo system segfaults every time I actually try to install something; I can browse the ebuilds and look at them, but not touch. Mandrake's graphical manager that I used somewhere around 7.0-ish would occasionally fail to segfault long enough to install a package or too, but it too would bomb out miserably, potentially corrupting files on the way out.

      I've never understood why something that you'd expect to be the linchpin of a user-friendly distribution has always failed so miserably and so downright reliably for me over the years, over all the computers, all the distros, and all the install options I've tried for them. Fortunately, there was dselect, apt-get, and now emerge which provide good enough text-based utilities.

      (I don't mind at all text-based utilities for installation. But graphical browsing of the available packages, by category, with a pane for the category, the description, the metadata, etc. is much nicer then anything you can do in text when you're just wandering around the distro seeing what is available.)
  • When I install a package, I want it to create all the icons and folders on the desktop and launch bar. And I want it to detect my window manager and adjust accordingly.
  • Why can't we have source tarballs that resolve dependencies?

    I want to be able to do "./configure && make && make install" and have the configure script look at a DEPENDENCIES file with URLs to resolve missing dependencies automatically. Such a system would be distro-independent and would not require packages to ever be created. If the process was this consistent, GUIs could even be created so the user doesn't even have to know what's going on.

    Just a thought from a guy that really hates package managers.
    • Such a system would be distro-independent and would not require packages to ever be created.

      No it wouldn't and yes it would, respectively. Distros have standards about where files go, and files in the wrong place can cause unexpected and difficult to diagnose failures. Moreover, if you don't notify the package system of these rogue programs your DEPEDENCIES files created, you're liable to end up with another instance of them on your system. In that case, the least of your problems is the wasted disk space; the odds of one or both of them stomping on each other's configuration files or binary compatibilities or something are extremely high, even for one package. Libraries could be a complete disaster here.

      As the AC sibling to this points out, you might consider Gentoo. Learn a little about how to twiddle ebuilds when you need to and you'll go a long way towards what you want, probably.
  • I really like slackware's approach...my only issue with it has been that it doesn't clean up after itself on an upgrade very well...
    Upgradepkg doesn't work as easily as it ought too, and when the distro changed to 8.x is actually got much worse because of the naming convention change.
    I am also impresses these days with Gentoo...and its ebuilds...the only thing I think they are missing there is the way they handle upgrades...while it think is great they don't wonk your current config files...I really think it would be better to add intelligence to merge the old config files and whatever new might need to be added during the emerge of an upgraded package.
    The only system I have used that I absolutely hated is the initial install of Debian...way to hard to use that system and totally not intuitive.
  • by mindslip ( 16677 ) on Friday January 03, 2003 @07:23PM (#5010523)
    1: Auto-xsu when trying to install/remove/update (such as the way red-carpet does)

    2: Auto-get dependancies off popular mirrors (like fr2.rpmfind.net which is far more up to date than www.rpmfind.net)

    3: Be able to list "--forced" and "--nodeps" packages, and remove / update / update dependancies of just those packages (i.e. clean up your system)

    4: Be able to list (-qa) with wildcards instead of having to -qa | grep whatever

    5: Standardize and *enforce* some sort of package topic structure, say based on Freshmeat's or Sourceforge's

    6: Be able to uninstall en-masse, any packages with no dependancies (i.e. clean up unused libraries, etc.)

    7: Stop this .deb / .mdk / .whatever-distribution madness and stick to the LSB or some such Linux standard... or at least auto-detect distribution and run appropriate script / install appropriate architecture files

    8: Curses interface for console (wouldn't "red-console" be nice?!?)

    9: Require packages to properly handle gnome/kde/etc. menus (Hey... even *windows* lets you *find* the stuff you install!!!)

    10: Be able to read and manipulate the packages installed on a system, when you've booted from a rescue disk (this is probably most important... booting off a rescue, then chroot'ing isn't always enough to get at the package databases, var directories, etc. db and other dependancies that RPM itself uses need to be *statically linked into RPM* so you can use it on a nearly-dead system.)

    11: Handle -bb and --rebuild better:
    11a: if you download a src.rpm file, and need to rebuild with a modified spec, you practically have to rebuild the src.rpm with the .spec edits.
    11b: Same for a tar.gz file with a spec in it... Edit the spec, re-tar the files.
    11c: Enforce proper .tar.gz filenames... it's quite common to find the tar filename different from what the .spec expects. Create a way to double-check the correctness of the spec
    11d: Auto-move/copy the tar.gz to /RPM/SOURCE ...I always do a build on a tar.gz and it says it can't find the tar.gz in /RPM/SOURCE! Well, heck, use the one I'm rebuilding and *move* it there!

    Ok, so hope these will help. I know they'd certainly help me!

    mindslip
    • 8: Curses interface for console (wouldn't "red-console" be nice?!?)

      It exists... there's a channel for Red Carpet in (where else?!) Red Carpet.

      You can now do pretty everything you did via the GUI, from command line. Sub/Unsub from channels, bring your system up to date, search for new packages to install, etc...

  • Rollback. (Score:3, Insightful)

    by Zapman ( 2662 ) on Friday January 03, 2003 @07:35PM (#5010618)
    People have some good points, but one thing that package managers need to understand is rollback. ie: I had gnome 1.4 installed. Worked well. I upgrade to gnome 2.1. 2.1 might just turn out to not do what I need at the moment, or it might be incompatable in an important way. I should be able to 'roll back' the system to 1.4 without performing major surgery.

    Yes, this will take more disk space (to hold the old versions of all the files, and the metadata to restore them). But in an enterprise environment, you need it. Sun has been doing this for years with their OS patches, and we should be able to steal it for packages too.
    • Re:Rollback. (Score:3, Interesting)

      by __past__ ( 542467 )
      Something similar: It would be nice if the package management system would remember what I only installed as a dependency of something else, and would remove it when I deinstall the other package (after asking me, of course).


  • For me the ultimate test was forcing some package installs and totally FUBARing my box, then managing to get to a console and doing an "apt-get -f install" and having apt-get magically fix all the things I broke. Thumbs up, Debian.
  • The thing that I like most about many Windows-based installers (these days) is the ability to step through the package install options and accept default values and have the package be installed in ready-to-run state.

    Mind you, I don't mean that the package silently installs to a pre-determined location automatically. Rather, it steps through the "usual" configuration values, but has sensible defaults for them such that a person doesn't need to override it under normal circumstances.

    It makes it easy for someone like my brother to install software without shooting himself in the foot.

    • One suggestion I'd add to this is making a *single* screen with all the config options on it, so that the user isn't going click, click, click, click through a series of dialog boxes.

      Then you slap an "OK" at the top.

      And, of course, retaining a non-interactive mode would be nice as well, where the defaults are simply used.
  • While RPMS and other types of packages may save some itme on slow computer with fast connections, I don't mind downloading source and building. Then I know it will work. When I was first introduced to FreeBSDs ports collection, I was only disappointed in that I didn't know about it sooner.

    Downloads the source, compiles, applies patches, and you are good to go. I never really had to uninstall much, but I think it worked well for that as well.

  • What's wrong with the System V package management tools?

    pkgadd/pkgrm/pkgtrans/pkginfo ...

    They work fine.

    What, you mean you don't like figuring out that SFWgcmn is the "Sun Freeware GNU Common Utilities" package?

    --NBVB
  • Here are some features that would be good.

    1. Revisioning. The ability to use different patch levels or versions. The ability to see how much space these different versions are taking up. Hard drive space is cheap, and I'm amazed an enterprise customer hasn't demanded this yet from Red Hat.

    2. Security Policy. RPMS and other package formats sometimes come with scripts to do necessary post-install stuff. It would be nice to have a seperate userspace library to handle the 'common' things that might be in these scripts like running ldconfig or mkfontdir. This way, it would be simple enough for the rpm tool to give you a rundown of what 'actions' needs to be done to install this application security-policy wise. Ideally most scripts would become superfluos and only used it really odd situations. And in those odd cases you can read through the script yourself.
    • Here are some features that would be good.

      1. Revisioning. The ability to use different patch levels or versions. The ability to see how much space these different versions are taking up. Hard drive space is cheap, and I'm amazed an enterprise customer hasn't demanded this yet from Red Hat.

      rpm's --repackage feature appears to have this intention behind it. Unfortunately, it doesn't appear to work here as the GPG and MD5 signatures aren't redone.

      2. Security Policy. RPMS and other package formats sometimes come with scripts to do necessary post-install stuff. It would be nice to have a seperate userspace library to handle the 'common' things that might be in these scripts like running ldconfig or mkfontdir. This way, it would be simple enough for the rpm tool to give you a rundown of what 'actions' needs to be done to install this application security-policy wise. Ideally most scripts would become superfluos and only used it really odd situations. And in those odd cases you can read through the script yourself.

      RH have begun to do this with things like chkfontpath, chkconfig, so I think they're aware of it...

      --

  • Beside many other things already mentioned I need a good auto/online update. Current SuSE allows me to keep all my packages up to date by 3-some clicks! I just don't have the time to keep up to date with all the security issues arising around the packages I have installed. If SuSE takes care about this I can perfectly live with it; installing their distro already means trusting them...
  • RPM and APT/DPKG depends on too many external tools.

    The package manager does most of the install/uninstall tasks by itself such as copying the file to the correct location, but to do the final install many packages depends on a preinstall and postintall scripts (and ditto for uninstall).

    The necessities of these scripts shows that the built in mechanisms of the package manager are not sufficient for all packages, so the package maintainer writes a script to add the missing pieces.

    These scripts can be a mess, usually written as a Bourne script or bash script, and depends on many tools such as sed, awk, perl, python, grep, ++++. Remove awk and many packages will fail during installation.

    To list all scripts on a RPM based system try:

    rpm -qa --triggers --scripts

    Almost every time I have seen a problem with a RPM package, it has been a bug or a missing tool in one of these scripts. I have been told that the same goes for DPKG (can someone confirm this ?)

    A better package manager should have very strong mechanisms for doing all sorts of things (such as installing an info page, removing old log files, or editing a configuration file). Look trough the scripts to get an idea of what is necessary. Then use one (and only one) script language without any external dependencies to write the remaining scripts (perl or python comes to mind, but an even purer solution would be to build in a small and powerful script language into the package manager).

  • by rafa ( 491 )
    Features I consider vital:

    One of the features I like about rpm is its --verify option, which checks if files in a package has been modified. Very useful - plus it informs you if a file is a configuration file - so those file changes can be ignored. verifying the integrity of a whole installation is just a small matter of a bit of awk to weed out the config files - then looking though the files that remains.

    Automatic checking of GPG signatures and md5sums - also vital.

    Freshening of packages, like the rpm -F, and query options that let you see what files belong to what package, and also ways to easily check what depends on what.

    Could be done better

    Better automatic dependancy resolution. apt-get for rpm is available, but neither it, nor mandrake's tools have worked well for me. red carpet looks promising, but I've never been able to use it with the latest mandrake versions properly (well, it's been a while since I checked last...

    Better LSB conformance, as mentioned by many here already.

    Better handling of -devel packages, automatic upgrading. Oh, and better source rpm handling - better dependancies for those, and they should have an option to install the built rpm!

    Options for interactive installs - where common options coudl be confirmed or changed at install itme - like debian (but preferrably only if I ask for them)

After the last of 16 mounting screws has been removed from an access cover, it will be discovered that the wrong access cover has been removed.

Working...