thanks for the detailed answer.
Let me say that my question did not in any ways imply that Pharo should
assume the burden of maintaining current images and VMs on old machines. My
goal was (is) simply to find out whether it was possible to repurpose a
number old machines I have laying around by installing *old* versions of
Pharo. And I now have a positive answer, having installed vm30 and verified
that it works perfectly.
In fact, I think adding a few lines to the website listing the last
supported version for the major MacOs releases would be a valuable addition
and nicely complement your commendable policy of keeping everything.
All the best,
On Tue, Mar 6, 2018 at 1:47 AM, Marcus Denker <marcus.den...@inria.fr>
> On 5 Mar 2018, at 20:16, stefano franchi <stefano.fran...@gmail.com>
> On Mon, Mar 5, 2018 at 11:28 AM, Stephane Ducasse <stepharo.s...@gmail.com
> > wrote:
>> Did you check
>> because we keep everything.
> Apologies for not having stated my question more precisely. Indeed, I
> started from get.pharo.org, but the zeroconf script did not guess right.
> The app it downloaded crashes at startup. Then I saw the *long* list of
> versions available and had no idea where to begin. So my real question is:
> Does anyone know which among the many VMs available on
> http://get.pharo.org would work on a MacOs Powerbook Pro running 10.6.9?
> It was the latest 32 bit only machine Apple made, based on the Intel Core
> Duo (NOT the Intel Core 2 Duo that came out a few months later).
> you could look for old VMs and downloads here: http://files.pharo.org/
> But it is quite hard to for us (with our limited man power) to support old
> machines forever… e.g the vm from that time should run,
> but at some point the VM gets improved and newer images require a newer VM
> as we want to actually use the features that
> new VMs provide.
> Keeping everything compatible in all possible directions (old images on
> new VMs, new image on old VMs …) puts quite some
> constraints on what you can do in future… an maintaining new VMs for all
> possible old MacOS versions could soon just
> use up all our manpower.
> So this is not a simple problem to solve. Even very financially capable
> projects (like Mozilla) can not support old MacOS
> versions. And they spend 150K per *month* just on CI infrastructure…
> imagine if they decide to not support anything older
> then MaOS 10.9… can we? should we?
> There are things to do on this front, but if I would spend effort the
> first thing I would work on is running *old* images on
> *new* VMs and explore what kind of abstraction would be needed to to that
> nicely and in a way that it can be maintained
> and in a way that all the needed code ( e.g. translation byte code from
> old to new) would be not part of the VM but
> part of the image.
> Making sure to run *current* images on old Machines can only be done by
> backporting the current VM to the old OS.
> This should be not that hard, worst case is that you need to combine some
> old OS related code with the rest of the new VM,
> but that should not be much.
> But one question: Considering what developer time costs… I am quite sure
> that it is cheaper to just buy a current Mac.