32768

[John Cremona]

I see that there are now over 2^32 trac tickets. Wow.  When I started computing you couldn't have integers greater than 32767...

John

[Samuel Lelievre]

2022-01-18 10:54:34 UTC, John Cremona:

>
> I see that there are now over 2^32 trac tickets.

The power of two that is near 32000 is really 2^15.

: )

> Wow. When I started computing you couldn't

> have integers greater than 32767...

Good thing we can now.

Here's to many more Sage tickets to come!   --Samuel

[John Cremona]

On Tue, 18 Jan 2022 at 13:53, Samuel Lelievre <samuel....@gmail.com> wrote:
>
> 2022-01-18 10:54:34 UTC, John Cremona:
> >
> > I see that there are now over 2^32 trac tickets.
>
> The power of two that is near 32000 is really 2^15.
> : )

You are right of course. We had 16-bit arithmetic so the max was
32767 (and min -32768). I still managed somehow to find the primes up
to 10^5 (using trial division, but I was only about 16, the programs
were on punched cards and I only got about one run a day which meant
that compiler syntax errors were very unwelcome.

I had better shut up. You can always watch the video from a Sage Days
in 2011 (https://wiki.sagemath.org/days29)!

John

>
> > Wow. When I started computing you couldn't
> > have integers greater than 32767...
>
> Good thing we can now.
>
> Here's to many more Sage tickets to come! --Samuel
>

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/3f822385-58a6-4438-9cb7-6fed2b2f436cn%40googlegroups.com.

[William Stein]

On Tue, Jan 18, 2022 at 7:25 AM John Cremona <john.c...@gmail.com> wrote:
>
> On Tue, 18 Jan 2022 at 13:53, Samuel Lelievre <samuel....@gmail.com> wrote:
> >
> > 2022-01-18 10:54:34 UTC, John Cremona:
> > >
> > > I see that there are now over 2^32 trac tickets.
> >
> > The power of two that is near 32000 is really 2^15.
> > : )
>
> You are right of course. We had 16-bit arithmetic so the max was
> 32767 (and min -32768). I still managed somehow to find the primes up
> to 10^5 (using trial division, but I was only about 16, the programs
> were on punched cards and I only got about one run a day which meant
> that compiler syntax errors were very unwelcome.
>
> I had better shut up.

Please don't!

Who will solve ticket #32768 -- https://trac.sagemath.org/ticket/32768

"centos-7-i386: SIGFPE while building dochtml"

with no other information at all.

My on topic question (given the power of 2 discussion) is:

I'm curious -- what is the situation is with Sage and 32-bit Linux?

I ask this because I recently wanted to build a specific Docker image
involving 32-bit pypy from source
in order to try out pypy,js, and couldn't do it! Support for "32 bit
linux distributions" has been shrunk a lot
in recent years ago (at least what was required by that Dockerfile),
in a way that really surprised me.

There's also an amazing project called v86, which runs 32-bit x86
operating systems entirely in your web
browser your Web Assembly.

https://copy.sh/v86/

(I think this is both amazing and at the same time completely useless
for probably anybody reading this.)

One of the difficulties for them is lack of support for 32-bit Linux
distros. E.g., if you click

https://copy.sh/v86/?profile=archlinux

you're running 32-bit Arch Linux in your web browser nearly instantly,
and it downloads files as needed
when you first reference them. But as they explain
https://github.com/copy/v86/blob/master/docs/archlinux.md
"The last ISO installer version of Archlinux that supports 32-bit is
2017.02.01. Later versions of the archis
os don't work on the v86 emulator because the installer only supports
x86_64, not x86 anymore. For existing
Archlinux installations, updates and patches will be done until
somewhere around 2018."

-- William


--
William (http://wstein.org)

[Matthias Koeppe]

On Tuesday, January 18, 2022 at 7:54:19 AM UTC-8 wst...@gmail.com wrote:

Who will solve ticket #32768 -- https://trac.sagemath.org/ticket/32768

"centos-7-i386: SIGFPE while building dochtml"

with no other information at all.

You can reproduce this failure by typing "tox -e docker-centos-7-i386-standard".

My on topic question (given the power of 2 discussion) is:

I'm curious -- what is the situation is with Sage and 32-bit Linux?

Supported, see list of platforms in https://wiki.sagemath.org/ReleaseTours/sage-9.5#Sources

as tested on each release tag on GH Actions. 

[William Stein]

Amazing.  Does anybody reading this use 32-bit Linux?

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/0fd7769f-fe6e-4d16-b5d2-59fb6164b6fdn%40googlegroups.com.

--

-- William Stein

[Michael Orlitzky]

On Tue, 2022-01-18 at 07:53 -0800, William Stein wrote:
>
> One of the difficulties for them is lack of support for 32-bit Linux
> distros. E.g., if you click
>

Nobody has the hardware any more. Debian is your best bet for ongoing
i686 support.

[William Stein]

Debian 32-bit is what pypyjs uses for its 32-bit build system (see
[1]), and even that seemed pretty broken when I tried a few months
ago.

In any case, trac ticket 2^15 is about a bug in 32-bit Linux Sage
support, and it's really impressive to me that you Sage devs fully
support 32-bit still :-).

[1] https://github.com/pypyjs/pypyjs/blob/4532320849881093635075db929240052300a844/tools/docker/build_base_image.sh


--
William (http://wstein.org)

[Dima Pasechnik]

By the way, 32-bit arm (still the standard for Raspberry Pi) is a bit
broken, something goes funny with weak references.
See https://trac.sagemath.org/ticket/28941

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CACLE5GBZUqcpbQa9OjSwp2vEV8HmRdAoo-CErkxRQfONrZyAOQ%40mail.gmail.com.

[Dima Pasechnik]

On Tue, 18 Jan 2022, 15:25 John Cremona, <john.c...@gmail.com> wrote:

On Tue, 18 Jan 2022 at 13:53, Samuel Lelievre <samuel....@gmail.com> wrote:
>
> 2022-01-18 10:54:34 UTC, John Cremona:
> >
> > I see that there are now over 2^32 trac tickets.
>
> The power of two that is near 32000 is really 2^15.
> : )

You are right of course.  We had 16-bit arithmetic so the max was
32767 (and min -32768).  I still managed somehow to find the primes up
to 10^5 (using trial division, but I was only about 16, the programs
were on punched cards and I only got about one run a day which meant
that compiler syntax errors were very unwelcome.

I also learned to program with punchcards. Various modern CI systems such as Travis CI and GitHub Actions

remind me of submitting jobs on punchcards, to be run on a clone of IBM OS360. Although IBM's JCL was way

worse than yml/yaml/toml used for CI configs :-)

 

I had better shut up.  You can always watch the video from a Sage Days
in 2011 (https://wiki.sagemath.org/days29)!

John

>
> > Wow. When I started computing you couldn't
> > have integers greater than 32767...
>
> Good thing we can now.
>
> Here's to many more Sage tickets to come!   --Samuel
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/3f822385-58a6-4438-9cb7-6fed2b2f436cn%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAD0p0K6JDR5Xm8TY1i1BdmEAaVTrN-Czmh9AB%2Bo1zB1m6MUDSg%40mail.gmail.com.

[Matthias Koeppe]

On Tuesday, January 18, 2022 at 8:32:47 AM UTC-8 Matthias Koeppe wrote:

On Tuesday, January 18, 2022 at 7:54:19 AM UTC-8 wst...@gmail.com wrote:

Who will solve ticket #32768 -- https://trac.sagemath.org/ticket/32768

"centos-7-i386: SIGFPE while building dochtml"

with no other information at all.

You can reproduce this failure by typing "tox -e docker-centos-7-i386-standard".

... or use one of the containers built by the CI - https://github.com/mkoeppe?tab=packages&q=centos-7-i386

[Thierry]

Hi,

On Tue, Jan 18, 2022 at 08:37:52AM -0800, William Stein wrote:
[...]

>
>
> Amazing. Does anybody reading this use 32-bit Linux?

Not using on a daily basis, but in in october 2021, i built a Sage
Debian Live 32 bit because Nicolas was going to teach Sage in Burkina
Faso (and i was teaching Sage in Algeria a couple of weeks later), and
to be honest, cygwin is too slow to be used during such a tutorial. The
image can still be downloaded at:

https://lipn.univ-paris13.fr/~monteil/hebergement/bobo/2021-10-16-sage_9.4-debian_bullseye-live_20210407.img
sha256sum : 7b6b9c6bc57bfd177286d20e7878e456aec67cda398d88bf3a43f291badbeaf9

Such Sage binaries are built within a qemu VM.

Note that not all optional packages are installed (some failed to
build), and some doctests are failing, which is why i did not release it
(and because i still want to fix some things in the SDL itself). But it
still contains most of Sage.

What seems clear to me is that, as i stopped to release SDL for a while,
the number of 32bit issues increased (by lack of pressure), also, most
upstream projects seem not test fo 32bit architecture anymore.

Apparently, there was at least one student with a 32bit laptop in
Burkina Faso, none in Algeria. The next SDL will be 64bit, though we
could continue to use my qemu machinery to build and test for 32bit
arcitecture.

Ciao,
Thierry

[Jan Groenewald]

Hi

I was interested in African universities and supporting 32 bit until about 2-3 years ago.

It is less than 1% of (science) students' laptops now, in my experience. And all institutional machines are 64bit.

Regards,

Jan

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/20220118191939.GA2005%40metelu.net.

--

  .~.
  /V\     Jan Groenewald
 /( )\    www.aims.ac.za
 ^^-^^ 

[Matthias Koeppe]

On Tuesday, January 18, 2022 at 11:19:44 AM UTC-8 Thierry (sage-googlesucks@xxx) wrote:

What seems clear to me is that [...] most

upstream projects seem not test fo 32bit architecture anymore.

Indeed many upstream projects have insufficient portability coverage even if they already use some CI. 

I have started to push some of our portability infrastructure upstream. See for example:

- https://github.com/linbox-team/fflas-ffpack/blob/master/.github/workflows/ci-sage.yml

- https://github.com/Normaliz/Normaliz/blob/master/.github/workflows/ci-sage.yml

- https://github.com/Singular/Singular/blob/spielwiese/.github/workflows/ci-sage.yml

- https://github.com/sympy/sympy/blob/master/.github/workflows/ci-sage.yml

I would recommend this approach when you run into portability issues with upstream projects.

[Volker Braun]

On Tuesday, January 18, 2022 at 4:54:19 PM UTC+1 William Stein wrote:

I'm curious -- what is the situation is with Sage and 32-bit Linux?

The Sage buildbot has 32-bit Debian 9, 10, 11 and it works. But using a recent (64-bit capable) CPU.

[seb....@gmail.com]

I have a 12 year old Acer AO Happy netbook (on Intel Atom CPU) which I'm running under LinuxMint. If I plan to do non CPU consuming work on the way to my office in subway trains and buses I still prefer it against a ThinkPad X240 because of weight, size and ergonomics. It still runs longer with its original battery as the ThinkPad with two batteries one of which I had to renew a year ago. In addition, for me its a matter of sustainability to use things as long as they work.

With Sage 9.4 I had to upgrade the LinuxMint from 17.3 to 19.3 which is the toplevel 32 bit LinuxMint. I hope that we will keep that for a while...

Building Sage from scratch nearly took a whole weekend. Currently I'm running Sage 9.5.beta3 on it with no problems.

[kcrisman]

In addition, for me its a matter of sustainability to use things as long as they work.

+1 

[Emmanuel Charpentier]

Le mercredi 19 janvier 2022 à 14:44:29 UTC+1, kcrisman a écrit :

In addition, for me its a matter of sustainability to use things as long as they work.

+1 

+1. And important in more ways than mathematical...