[Live-demo] Rethinking osgeo-live
tech_dev at wildintellect.com
Thu Oct 25 14:22:19 PDT 2012
Some other candidates include...
Juju (new almost identical thing for Ubuntu specifically)
On 10/25/2012 02:07 PM, Cameron Shorter wrote:
> An interesting idea you have brought up.
> Here are some thoughts that Jody Garnett mentioned in IRC today...
> <jgarnett> there is a slightly different approach to consider
> <jgarnett> if you look at things like "home brew"
> <kalxas> but I don't believe it fits the way GNU/Linux works
> <jgarnett> they are very similar to what osgeo live does
> <jgarnett> osgeo live has a bunch of tested install scripts
> <jgarnett> that "grab what is needed"
> <jgarnett> we have caught developers just running the scripts on their
> linux machine
> <jgarnett> (and indeed they sometimes work for mac as well)
> <jgarnett> home brew, and its ports to linux, are similar
> <jgarnett> would turn the project into a set of install scripts
> <CameronShorter> jgarnett, got a url for home brew?
> <jgarnett> the "live demo" would be a downstream project; where the
> install scripts have been applied to a ISO
> <jgarnett> or to a VM
> <jgarnett> http://mxcl.github.com/homebrew/
> <sigabrt> Title: Homebrew — MacPorts driving you to drink? Try Homebrew!
> (at mxcl.github.com)
> <jgarnett> (it is not so much this specific "home-brew" project - it is
> the approach that is similar to the core of osgeo live)
> <jgarnett> (i.e. the part we ask projects to maintain)
> <jgarnett> but yeah - was not going to wade into that email debate
> <jgarnett> (here is a port of homebrew to linux
> http://blog.frameos.org/2010/11/10/mac-homebrew-ported-to-linux/ )
> <sigabrt> Title: Mac Homebrew ported to Linux - Automation Inc. (at
> <kalxas> "Caveats * The port is still a work in progress * Most of the
> existing haven’t been tested. * Port has been tested only in Ubuntu and
> On 25/10/2012 7:26 PM, Angelos Tzotsos wrote:
>> And something I forgot to write down:
>> What about security updates? We would need to maintain libraries like
>> Qt, while this is done upstream right now.
>> On 10/25/2012 11:16 AM, Angelos Tzotsos wrote:
>>> Hi Barry,
>>> On 10/25/2012 10:29 AM, Barry Rowlingson wrote:
>>>> On 25 October 2012 03:33, Brian Hamlin <maplabs at light42.com> wrote:
>>>>> this is breathtakingly unrealistic :-)
>>>> Thanks! :)
>>>> I'll approach some of the criticisms...
>>>> * Alex, yes, correctly compiling Windows things is hard. But someone
>>>> has done it for OSGeo4W, Jo has done it for Portable GIS. That
>>>> expertise exists.
>>>> * Alex, Angelos: this wouldn't be statically linked binaries
>>>> (HUUUUGE!) but more like a python virtual env. There would be some
>>>> wrapper to make sure qgis always links with /osgeo/lib/libqt4.so, and
>>>> never with /usr/lib/libqt4.so. Essentially it would be everything that
>>>> is currently on the live disc (except the kernel and user tools), with
>>>> the applications configured to get their dependencies from the right
>>> I see what you mean, I am not sure it is wise to bypass
>>> distributions. How many software providers do this today?
>>> Even Google with all those resources available has not been able to
>>> provide one binary for all Linux machines...
>>> We would need to re-invent the wheel in some cases to do this.
>>> But this does not mean I don't like the way you are thinking.
>>>> * Angelos: I demand (and get) freedom for my desktop, at the cost of
>>>> not having central IT back my machine up. However, if I take a live
>>>> DVD to a Windows-using, central-IT supported colleagues desk and go
>>>> 'hey, look at this', all I get is frustration and eventually a BIOS
>>>> password prompt. So I then have to go back with OSGeo4W.
>>> Unfortunately freedom is still an every-day battle.
>>> At least we are winning on the server side.
>>>> * Angelos: running in your favourite OS should be as simple as
>>>> copying the things you want to run to your currently existing and
>>>> configured-exactly-how-you-like-it OS.
>>> I know packaging is not perfect today, but is much better than it
>>> used to be 5 or 10 years back.
>>>> I don't see the point in making
>>>> an openSUSE version - we're trying to promote OSGeo s/w here, not
>>>> GNU/Linux distributions.
>>> This is exactly why this version is not available...
>>>> I'm just thinking that there's more worth in concentrating efforts in
>>>> getting OSGeo applications out there rather than spinning up new
>>>> Ubuntu distributions every six months. To that end, a simple,
>>>> user-driven binary installation process would seem to be optimal. We
>>>> have OSGeo4W, why not OSGeo4L and OSGeo4M?
>>> Because GNU/Linux is all about freedom of choice. No matter how we
>>> decide to create packages, people will want native packages for their
>>> distribution through UbuntuGIS, DebianGIS, EL, OBS, AUR etc.
>>> I agree that OSGeo4W is a huge project that fits exactly the needs of
>>> Windows users. I am sure a OSGeo4Mac would also be very successive.
>>>> I don't see a technical barrier to this, so it's just limited by
>>>> resources (our time!). Anyone got a spare 20 years?
>>> If we had the spare time and I was to make this decision, I would
>>> prefer offering deb and rpm files for all OSGeo related projects and
>>> I would be left with 10 years for vacations :)
>>>> Live-demo mailing list
>>>> Live-demo at lists.osgeo.org
More information about the Osgeolive