Some more observations in this thread: Well I have been reading up a bunch on the Apt vs Portage stuff and it is validating my inclinations to go with Apt more. Really when one says Portage is a lot more flexible and powerful, it is so because of it's source build nature. The whole idea behind Portage is to allow an insane level of customization via custom options and tuned source builds. Yes you "can" install binary blobs via Portage but that is not it's central idea. It is not only about what you "can" do with a system, it is also about what the system is "designed" for. In case of Portage it is build from source. Using a variation of the theme to install binary blobs via ebuilds does not appear appealing. Much of Portage's power will be of no meaning if one is exclusively using binary packaging. In addition since that is not the central idea we do not know how well that option works, what kind of surprises are there.
>From whatever I have read Portage and by implication Gentoo is much more of a hacker's toolbox requiring one to dig deep into technical nitty gritties - not quite an user-friendly hands off thingy. That is not to say that it is bad but it serves a specific purpose. In addition there are better GUI's for Apt since Portage by it's very nature is not very GUI friendly. The following is an excellent article on this: http://www.tomvergote.be/writings/Linux/Debian-Gentoo-production-environment.html It talks in general about Gentoo vs Debian but a majority of the points revolve around packaging. I have had very good experiences with Apt in the past. It is robust, mature and rock-solid. I have faced multiple UPS failures switching off a system multiple times in the middle of a dist-upgrade and every time Apt will recover from the exact point at which things stopped and finally gave me a working system. Apt has quality GUIs for it like Synaptic or PackageKit etc. that can be ported and used. I see a lot of value in moving to the Apt/Dpkg port (unless of course some silly surprise like what popped up for pkg-get happens). Regards, Moinak. On Fri, Aug 29, 2008 at 11:34 AM, Moinak Ghosh <moinakg at belenix.org> wrote: > Having said all this. I'd appreciate any pointers to Emerge/Apt comparison. > > Regards, > Moinak. > > On Fri, Aug 29, 2008 at 11:32 AM, Moinak Ghosh <moinakg at belenix.org> wrote: >> On Fri, Aug 29, 2008 at 9:42 AM, russiane39 at gmail.com <mail at >> belenix.org> wrote: >>> Heh, again everyone will blame me, but why we can't use gentoo emerge for >>> Belenix ? Most of work is done, I was able to successfully bootstrap >>> emerge-based system on Belenix 0.7.1 . Emerge will give us most flexibilty >>> and ease of use. Here is links for further investigations: >>> Old (probably abandoned) Alba experiment - >>> http://gentoo-wiki.com/Alba_Experiment >>> Maintained Gentoo Prefix for Solaris - >>> http://www.gentoo.org/proj/en/gentoo-alt/prefix/bootstrap-solaris.xml >> >> Suggestions are always welcome. However I personally dislike emerge. >> I do not like my box spending days compiling stuff just to upgrade. In >> addition the OpenSolaris software ecosystem is kind of unique in that >> it is divided up into different consolidations with each using it's own >> build >> mechanism. It is not worthwhile or practical to try and unifiy all those >> disparate build mechanisms to the Emerge framework. What is needed >> is a purely Binary packaging setup separate from the build systems. In >> this I agree with the viewpoints of the IPS folks. >> >> While Emerge can handle binary packages it's main value is in tuning >> builds for the platform. In addition it is more work to integrate emerge >> with the OpenSolaris features like ZFS and Zones. >> So I still believe the Nexenta APT/Dpkg mechanism is extremely robust >> and an excellent port. Biggest of all, almost all of the heavy lifting has >> already been done, it is already tuned for utilizing ZFS and supporting >> Zones. All that needs to be done is to make it work properly with the >> Snap Upgrade framework. >> >> Even though this is a longer term change and not to be done right now >> I am at the verge of putting the lid on IPS. It is an attempt to re-invent >> a >> very large wheel and I do not agree to some of the design choices among >> other problems. I have read through most of the observations/reasoning >> around IPS and do not see any point that is not addressed by the major >> existing mainstream packaging systems. >> >> Regards, >> Moinak, >> >> -- >> ================================ >> http://www.belenix.org/ >> http://moinakg.wordpress.com/ >> > > > > -- > ================================ > http://www.belenix.org/ > http://moinakg.wordpress.com/ > -- ================================ http://www.belenix.org/ http://moinakg.wordpress.com/
