On 10 February 2015 at 17:42, mpb <mpb.m...@gmail.com> wrote: >> On 9 February 2015 at 17:18, Sergio Tortosa Benedito <serto...@gmail.com> >> wrote: >>> Right now we rely on Install and Compile, but I've been thinking about >>> changing the package management way. Lately, there's been a shift towards >>> bundles, and I find Gobo a perfect fit for this. I thought we could use a >>> bundling system called Limba for applications and a something like OStree >>> for the main OS. What do you think? >>> PD: I'm aware this is thinking big but it's something should be defined. > > On Tue, Feb 10, 2015 at 9:42 AM, Liam Proven <lpro...@gmail.com> wrote: >> I had to go and Google these projects, but now that I have a very >> vague general understanding of what they do, I see your point. >> >> However, this is proposing very major changes to a distro that is >> already critically under-staffed at the moment anyway. >> >> Might I suggest that you write up 2 or 3 detailed descriptions of what >> both of these things would mean, separately and jointly, explaining >> what Limba and OSTree are, what benefits they would bring, what costs, >> etc.? That way people could judge for themselves without having to go >> and do possibly hours of research. > > Here are the links I found when searching for Limba and OSTree. > > http://blog.tenstral.net/2014/11/introducing-limba-a-software-installer-experiment.html > > https://wiki.gnome.org/Projects/OSTree > > The Limba post references this "famous blogpost" by Lennart Poettering > from September 2014. > > http://0pointer.net/blog/revisiting-how-we-put-together-linux-systems.html > > After a cursory review, it seems to me Poettering's proposal offers > solutions to many of the problems that motivated the creation of Gobo. > Given Poettering's influence and record of accomplishment, I'd > recommend the Gobo devs take a look at it.
I read this when it first came out. The similarities are obvious. I don't remember the exact details now, but I remember that upon first reading I spotted some loose stitches in his proposal. IIRC, some of the things were properly namespaced, but some were not, so while it seemed like a modularized system where you could mix-and-match bits and pieces from different distros, in practice you would still have ABI dependencies on some core frameworks, so his idea that upstream devs would be able to offer their own per-project binaries breaks at the moment you're depending on a core framework that's distro-specific. I may be mistaken since it's been a while since I read that post (and I don't have the time to revisit it now, I'm writing this as I wait for a compile to finish :) ), but my memory of reading it is that it had some ideas that mostly sound like the concept of Frameworks from Mac OS X, but it wasn't really a holy grail of modularity and interoperability for Linux. -- Hisham _______________________________________________ gobolinux-devel mailing list gobolinux-devel@lists.gobolinux.org http://lists.gobolinux.org/mailman/listinfo/gobolinux-devel
