1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

PBIs coming to 9.0

Discussion in 'Porting New Software' started by noz, Oct 29, 2010.

  1. noz

    noz New Member

    Messages:
    41
    Thanks Received:
    11
    I just read this in the quarterly status report and was wondering if anyone else is excited about this as I am. Maybe it will completely replace packages.
     
  2. UNIXgod

    UNIXgod New Member

    Messages:
    1,089
    Thanks Received:
    195
    Is that the goal? replace pkg_add?

    I have never used PC-BSD. I though the goal was to ubuntu{ize} FreeBSD to simplify desktop installs for new user experience. Not displace it.
     
  3. noz

    noz New Member

    Messages:
    41
    Thanks Received:
    11
    No, it was just speculation on my part.

    I've only used PC-BSD a few times. I never stuck with it, but loved the idea of PBIs; a single self-contained installer that frees the user from any dependency worries. I used packages when I first started fiddling with FreeBSD and always got in trouble when I mixed packages and ports (I only stick to ports now).
     
  4. Beastie

    Beastie Active Member

    Messages:
    1,941
    Thanks Received:
    343
    Read it yesterday.

    ... and turns a minimalistic setup into a fat bloated crappy setup requiring 1.5 times as much disk space or so. Give me dependency "hell" any time of the day. There goes my excitement about it.

    Anyway, I hope it never gets adopted as the default packaging system.
     
    vertexSymphony thanks for this.
  5. fronclynne

    fronclynne New Member

    Messages:
    1,297
    Thanks Received:
    168
    I use ports(7) so my opinion shouldn't be too weighty, however pkg_add(1) (as synecdoche for the package system as differentiated from compiling from ports) could merely be extended to deal with a series of statically compiled packages. Maybe even as simply as the already provided shells/bash-static.

    Anyway, if the excellent droogies @ PC-BSD wish to provide the packaging system (as they already do) and the packages themselves (as they already do) I can't see having it as an option as a bad thing. I think doing away with the already excellent pkg_add(1) in preference for PBIs would be silly.
     
  6. noz

    noz New Member

    Messages:
    41
    Thanks Received:
    11

    If you're a person that likes to keep your system as lean as possible I can respect that. There are also those that would like to use FreeBSD as a desktop OS that would gladly sacrifice a little HD space for convenience.

    Like I said, I was just speculating. It's much more likely that PBIs would just become one more way of installing software rather than it replacing anything.
     
  7. Ralph_Ellis

    Ralph_Ellis New Member

    Messages:
    28
    Thanks Received:
    5
    PC-BSD is really just adding a third way to install packages where some of the configuration options are already sorted out in advance. If you want to control all of the compile choices from the start, you are better off with the ports system. PBIs are a very convenient and simple way to add complex packages. For example, the PBI for Firefox automatically installs Flash also without you having to link the browser to the linux binary. The PBI for Wine will allow you to install Wine on a 64 bit system without you manually having to set up aliases to the 32 bit binaries.
    PBIs are a genuine time saver and they are not issued without being tested for bugs. They are also linked to the ports that are involved so that they will update and you can be notified about updates.
    The PC-BSD community is really just giving back to the FreeBSD community. In the end, PC-BSD is 100% FreeBSD with desktop oriented customizations.
     
  8. oliverh

    oliverh New Member

    Messages:
    557
    Thanks Received:
    38
    It's a nice flavour of FreeBSD for a certain kind of desktop-user, it's maybe an alternative, but it's definitely not an alternative for something I call K.I.S.S. -- keep it simple stupid. Well, that's my opinion and as long as it gets out of my way I'm happy. The idea of PC-BSD itself is great, I like their commitment, otherwise the idea of PBI's is crap in my opinion. We are seeing the rise of ARM on the desktop and the server, smaller systems, different from the crappy x86-architecture. There you have to think again about the terminus technicus "resource-hog".
     
  9. Ralph_Ellis

    Ralph_Ellis New Member

    Messages:
    28
    Thanks Received:
    5
    In the end, it is all about choice. A PC-BSD system can be as light as a FreeBSD system simply by the choice of programs, desktop environment and system settings. I have a PC-BSD 8.1 64bit system with ZFS on one hard drive and I have a FreeBSD 9-current 64bit system with UFS on another. I wanted a system where I could start by building everything from source and customize it. I did however import the majority of configuration files from PC-BSD. They save a lot of time.
    The PBIs are really a bundled packages system with the configuration choices already made. No different or worse than using packages. If people do want a faster, lighter, more customizable system then you need to build from source, choose your programs carefully and customize your kernel.
    To each their own.
     
  10. sk8harddiefast

    sk8harddiefast Active Member

    Messages:
    2,095
    Thanks Received:
    167
    I use only ports and I love flags. I don't want to use PBI or any other GUI installer :( "Make config & Make install clean" do the job just fine.
    And I present you... PCBSD :p
    IMO PCBSD created because some people wanted an more....Windowing BSD system with easy and fast setup. There is no reason to import PCBSD's stuff on FreeBSD (except if is something very important like ZFS option on sysinstall) because this is the reason that PCBSD created. To be a Desktop FreeBSD OS. And FreeBSD is FreeBSD and not PCBSD. Are the 2 sides of the same coin and every side exists for some reason and to serve a specific purpose.
     
  11. Beastie

    Beastie Active Member

    Messages:
    1,941
    Thanks Received:
    343
    I have all my (system) configuration files backed up. All it takes to restore them is a tax -x and a few cp commands (and some modifications if the hardware is different). It is all done within minutes at most.

    To each their own, yes. But FreeBSD's .tbz packages =! PC-BSD's .pbi packages. One may still want to install FreeBSD's packages (not ports) instead of the statically-linked 1-package-per-application PBIs. Packages are fast to install compared to ports and smaller than PBIs so in certain cases they are the best choice.
    So, as long as the two existing alternatives are kept in the base system, I have no problem with a third one. I have a problem with it if it eventually replaces one of them.


    ~~~~


    My understanding of the sentence
    is that it supports text-based installations from the shell, just like pkg_add and make install.
    Still, the idea of statically-linked packages does not appeal to me at all.
     
  12. FreeMWP

    FreeMWP New Member

    Messages:
    28
    Thanks Received:
    4
    I am testing the new PBI format on a clean FreeBSD installation right now. The new PBI format now supports:

    Testing on a clean FreeBSD can be done by:

    Code:
    # svn co svn://svn.pcbsd.org/pcbsd/current/src-sh/pbi-manager
    # cd pbi-manager
    # ./install.sh


    For now, there are no pbi repo, but you can make your owen from ports using the pbi_makeport tool.

    For more info check http://wiki.pcbsd.org/index.php/PC-BSD_9_Handbook
     
  13. Jamz

    Jamz New Member

    Messages:
    16
    Thanks Received:
    2
    When ZFS v28 arrives with de-dupe the extra dependency bloat from PBIs should be minimised.

    James
     
  14. ckester

    ckester New Member

    Messages:
    288
    Thanks Received:
    38
    I know next to nothing about pbi's, but I have to ask: what do people who like that sort of thing expect to get from having them on FreeBSD that they can't already get on PC-BSD?

    If there's something else on FreeBSD that they miss on PC-BSD, wouldn't it make more sense to migrate those bits there?

    Why the continual push to make FreeBSD more "user friendly" when PC-BSD is already addressing that need?

    (Like some of the others who have already commented, I use ports only and never packages. So I don't really care if pbi's replace pkg's or simply exist alongside them. But I wonder where all this is going, and I'm not sure I like it.)
     
  15. Jamz

    Jamz New Member

    Messages:
    16
    Thanks Received:
    2
    I'm not sure what you're worried about, it's just another app (pbi-manager) available in the ports tree, you're under no obligation to use it..
     
    noz thanks for this.
  16. oliverh

    oliverh New Member

    Messages:
    557
    Thanks Received:
    38
    When ZFS v28 arrives, it will be certainly 9.1 or 9.2. ZFS is bloat and bloat plus bloat is equal to more bloat. The consumer-desktop, you're referring to, is a fraction of a fraction in UNIX-like environments. The main use is: servers, embedded systems and scientific workstations. The requirements on those platforms are different: stability and reliability. UFS is still the number one if it comes to stability and reliability, most of the time it's just poorly configured in terms of performance. Counting version numbers and abandon K.I.S.S. in favour of bling-bling for the consumer-desktops leads straight to Windows-like disaster-prone environments. If you want so see an example, have a look at Ubuntu, Fedora and OpenSuse.
     
  17. noz

    noz New Member

    Messages:
    41
    Thanks Received:
    11
    Why not? Do you think it's a bad thing? Should FreeBSD move in the opposite direction and become as user unfriendly as possible? Should FreeBSD not move in any direction at all?
     
  18. ckester

    ckester New Member

    Messages:
    288
    Thanks Received:
    38
    If you would try to understand why I put "user friendly" in snicker quotes, perhaps you'll understand why I would also say, "With respect to this, I don't see why FreeBSD should move in any direction at all."

    I just don't get it, and I wish someone would answer my question about why they often seem to want to make FreeBSD indistinguishable from PC-BSD. (Or worse, Ubuntu.)
     
  19. DutchDaemon

    DutchDaemon Administrator Staff Member Administrator Moderator

    Messages:
    10,822
    Thanks Received:
    1,923
    FreeBSD appears to be very 'advanced-user-friendly', especially to people using it on servers and occasionally on a desk/laptop. Perhaps moving too much in the direction of 'beginning-user-friendly', where almost everything is graphical and/or decided for you, is not the way in which a majority of current users want to move, because it might take away from the tools that get the job done excellently right now. Indeed: PC-BSD is the way to go if that's what people want. No need to duplicate (or port over) these efforts to water down FreeBSD. If you get bored with PC-BSD eventually, FreeBSD is the next step. You'll have some hands-on experience by then.
     
    ckester thanks for this.
  20. noz

    noz New Member

    Messages:
    41
    Thanks Received:
    11
    Because no one knows where your question is coming from. Just because FreeBSD is getting the option to use PBIs, all of a sudden it's indistinguishable from PC-BSD? If anything, at least one component of PC-BSD (PBIs for instance) is becoming more like it came from FreeBSD because it's been completely rewritten as a commandline tool instead of being GUI only.


    All the GUI elements in FreeBSD are optional and off by default. I don't think that's changing any time soon, and there's no indication of it becoming otherwise. Fear and paranoia about GUIs and "bling-bling" taking over FreeBSD is unsubstantiated.
     
  21. oliverh

    oliverh New Member

    Messages:
    557
    Thanks Received:
    38

    Working since Irix and Siemens SINIX in UNIX environments I rather call this experience, than paranoia. Just for your information, I was a member of the DesktopBSD team ... you know, this flavour of FreeBSD for the user-friendly desktop -- different approach than PC-BSD. Apart from that, read DutchDaemons comment:

    See? It's that easy.
     
    ckester thanks for this.
  22. ollopa

    ollopa New Member

    Messages:
    1
    Thanks Received:
    4
    As was already half-explained, the latest PBI system uses hashes to identify files that are common to packages and hardlinks to share them without bloating the system. You're really not losing anything when compared to the current packages system but you're gaining quite a bit in terms of ease of management. It can be very useful in some environments like, for example, if you need to keep two pieces of software with conflicting dependencies. Sure there are disadvantages to binaries and other ways to solve the same problems, but actually the PBI system works really well and surely deserves more credit than some people around here are giving it.

    I really don't understand the hostile reaction to the notion of migrating some PC-BSD features to FreeBSD. Nobody is trying to GUI up FreeBSD or water it down. The PC-BSD team are very good people and they are not dumb. The solutions they come up with are well thought-out and they act quickly to mitigate the weaknesses. Not everything they come up with is strictly for an easy user experience or a flashy graphical front-end. Have a go someday at installing FreeBSD using PC-BSD's pc-sysinstall, for example--and I don't mean via the gui front-end, but from the command-line with a hand-built config file. It's pretty cool.

    Anyway... I've been a FreeBSD admin for about 10 years now and mostly use ports. I've also used PC-BSD and PBIs, I'm familiar with the concept from GoboLinux and I've created a PBI before and talked with Kris Moore about the changes in PBI9. Trust me you have nothing to fear from gaining PBIs or anything else from the PC-BSD guys. FreeBSD is being enhanced, not watered-down.
     
    cuq, kipz, swa and 1 other person thank for this.