On Mon, Jun 19, 2023 at 03:46:34PM +0100, Dima Pasechnik wrote:
> On Mon, Jun 19, 2023 at 3:05 PM Waldek Hebisch
> <[email protected]> wrote:
> >
> > On Mon, Jun 19, 2023 at 10:29:44AM +0100, Dima Pasechnik wrote:
> > > On Mon, Jun 19, 2023 at 8:44 AM Ralf Hemmecke <[email protected]> wrote:
> > > >
> > > > On 19.06.23 09:22, Grégory Vanuxem wrote:
> > > > > I wonder why you're using such an old LISP?
> > > >
> > > > Waldek wants to make sure that jfricas also works with the sbcl that he
> > > > puts into the binary fricas release. In fricas-1.3.8 this was
> > > > sbcl-1.1.1. So I presume that I should test 1.1.1.
> > >
> > > it's wishful thinking that SBCL 1.1.1.1, released in 2012, will
> > > correctly work on all modern OSs.
> > > Waste of time, if you ask me...
> >
> > Well, download binary 1.3.8 (based on sbcl 1.1.1) and try it. It
> > works on all systems available to me and up to now I had no report
> > of failure due to incompatibility with the system.
>
> A system not certified by its makers to work on an OS/hardware,
> because it's way too old,
> should be avoided, full stop.
Burden of ensuring that existing software works in on OS and
hardware makers. Linux actually delivers quite good compatibility.
I agree that software which does not take compatibility seroiusly
should be avoided. Unfortunately, it looks that ASDF seriously
messed compatibility and signifcant portion of Lisp world
depends on ASDF.
> As well, I imagine that code compiled, for a modern CPU, with an up to
> date sbcl, will be considerably
> faster than the one built with sbcl 1.1.1, as the latter has no idea
> about modern CPUs.
Well, this is more about optimizations present in compiler. And
recent sbcl optimizes better. Compilers are reasonably mature
business and few percent improvement is "considerable", but
such improvements are frequently sacrified for partablity.
> > Of course,
> > at some time we will have to move to newer version. But using
> > newest version is essentially warranted to fail on some systems,
> > so we need to choose correctly.
> >
> > More generaly, if you declare everything more than 3.5 years as
> > obsolete and decide to break it, that is your right. But not all
> > folks adapt such point of view and it is possible to create
> > binaries compatible with wide range of system versions.
>
> Why do you want to encourage these folks to stay on obsolete OSs, which
> have reached their EOL, are insecure and buggy?
> You want them to get hacked? :-)
Well, unfortunately, installing software is turning into constraint
satisfaction problem, there are many constraints and user have to
find combination of versions that satisfies all constraints.
Much of that is not needed at all or could be much less
restrictive. And I do not want to add extra constraints to
FriCAS use.
Concerning security, normal life cycle for software is like
follows. In first few years after release there is a bug fest,
when bugs in new functionality are discoverd and fixed.
Then software enter stable period, with few bugs found.
After several years things start to break up because
somebody broke them. Security problem are due to bugs,
so if you want most security you should use stable versions.
Most major Linux distributions provide security updates
for 10 years, in this period old software is likely to
be more secure than new one.
Most security threats came from network, so old system
(even without security updates) not connected to network
is likely to be more secure than system connected to the
net. Yet people keep connecting computers to the net.
Anyway, there are legitimate reasons to run really obsolete
systems and I have no principal objection to people doing
this. Of course, there are limits: if some people want
to run FriCAS on real 370, they will have my sympathy,
but probably no real help. But I consider 10 years minimal
reasonable support window.
--
Waldek Hebisch
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/fricas-devel/ZJc%2BMIjNoxW/E4Wg%40fricas.math.uni.wroc.pl.
