[OSGeoLive] Will there be an ARM version of the osgeo live?

Kobben, Barend (UT-ITC) b.j.kobben at utwente.nl
Mon Feb 7 07:58:31 PST 2022


You are correct in being “a bit surprised the M1 Macs can’t run x86 OSes in emulation”, but unfortunately the emulation (using a thinbg called Rosetta)  is working for apps, not for OS-es (such as OSGEO Live Lubuntu) as a whole. From the site
https://www.computerworld.com/article/3597949/everything-you-need-to-know-about-rosetta-2-on-apple-silicon-macs.html : “What can’t Rosetta 2 translate? Rosetta cannot translate kernel extensions or Virtual Machine apps that virtualize x86_64 computer platforms. Developers should be aware that Rosetta is also unable to translate AVX, AVX2, and AVX512 vector instructions. ”

Barend Köbben

On 07/02/2022, 15:48, "osgeolive on behalf of James Klassen" <osgeolive-bounces at lists.osgeo.org<mailto:osgeolive-bounces at lists.osgeo.org> on behalf of jklassen at sharedgeo.org<mailto:jklassen at sharedgeo.org>> wrote:

There was discussion awhile back about supporting ARM for Raspberry Pi and similar SBCs that came to the same conclusion that it would take more developer resources that were available.

OSGeo Live is meant to “just work” to encourage new users explore the software without having to first face the learning curve of getting it installed and configured correctly.  That is a lot more difficult to accomplish when users face to face the variations inherent in running different architectures.

Most, but not all of the packages that go into OSGeo Live are available on ARM (are in Ubuntu-GIS and Debian-GIS or are platform agnostic and install the same files as on x86).  So, technically it isn’t too far fetched.  But, if I remember correctly, pain points are testing and documentation.  I’d venture a guess that, by far, nearly all of the developer time on OSGeo Live is spent on testing and documentation.

Another issue with ARM is that while the user space is the same/similar across ARM devices, a bootable image (like we do with x86) would have to be tailored to each device.  Maybe there would be a way to just provide a user space and have the user provide the matching version of Ubuntu for their machine.  Maybe the whole thing could be built into a snap or flatpak or appimage.  It would still be a different experience than we’ve traditionally had for x86 which raises documentation and ease of use concerns.

I’m also a bit surprised the M1 Macs can’t run x86 OSes in emulation.  There were programs that emulated a PC to allow 68k and PowerPC  era Macs to run DOS/Windows.

On Sat, Feb 5, 2022 at 04:06 Angelos Tzotsos <gcpp.kalxas at gmail.com<mailto:gcpp.kalxas at gmail.com>> wrote:
Dear Barend,

We do not have an ARM version. This would require more developer
resources than we currently have, so there is currently no plan to
support this architecture.


On 2/2/22 01:24, Kobben, Barend (UT-ITC) wrote:
> For installation in the Parallels virtual machine on a new MacPro (running on the Apple silicon architecture), an ARM version instead of an Intel version is needed. Is that available, or will in be...? Or are there alternative ways to get it running on a Mac M1...?
> --
> Barend Köbben
> Senior Lecturer – ITC-GIP & ATLAS, University Twente
> PO Box 217, 7500 AE Enschede (The Netherlands)
> +31-(0)53 4874 253 / room 1-065 ITC
> _______________________________________________
> osgeolive mailing list
> osgeolive at lists.osgeo.org<mailto:osgeolive at lists.osgeo.org>
> https://lists.osgeo.org/mailman/listinfo/osgeolive

Angelos Tzotsos, PhD
Open Source Geospatial Foundation

osgeolive mailing list
osgeolive at lists.osgeo.org<mailto:osgeolive at lists.osgeo.org>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/osgeolive/attachments/20220207/ee18fcb4/attachment.html>

More information about the osgeolive mailing list