Proposal: Stop providing binary distributions

[Matthias Koeppe]

Comments/review please in https://trac.sagemath.org/ticket/33131

[Emmanuel Charpentier]

Mixed feelings... See my comments on the ticket.

[Murray Eisenberg]

As strictly an end-user of Sage, and only on macOS, I find the prospect of not being able to download a binary (not even from https://github.com/3-manifolds/Sage_macOS/releases) repellent. And providing, only a Docker image instead, equally repellent. The probability is large that I would abandon Sage if binaries were not available.

[William Stein]

On Sun, Jan 9, 2022 at 12:46 PM Murray Eisenberg
<murraye...@gmail.com> wrote:
> As strictly an end-user of Sage, and only on macOS, I find the prospect of not being able to download a binary (not even from https://github.com/3-manifolds/Sage_macOS/releases) repellent. And providing, only a Docker image instead, equally repellent. The probability is large that I would abandon Sage if binaries were not available.

I was also really disturbed when I saw the subject of this post. However,
upon closer inspection, I suspect that the subject "Proposal: Stop
providing binary distributions" of this
thread requires clarification. The actual proposal at the trac ticket is
"Installation manual: Remove mention of Sage-mirror-hosted binary
distributions", and to instead explicitly point to several (often
better) sources of binary distributions. For example,

https://github.com/3-manifolds/Sage_macOS/releases

is NOT a "sage-mirror-hosted binary distribution". I think what the
3-manifolds people
provide is exactly the sort of third party binary distribution of Sage
that Matthias is proposing
to *better* advertise, highlight and emphasize in the Sage docs.

Matthias, correct me if I'm wrong, but are you proposing that we just a
better job at pointing users to binary distributions of Sage such as
https://github.com/3-manifolds/Sage_macOS/releases, conda-forge, and
of course any packaging of Sage for Linux distros, etc., instead of
pointing them at binaries that don't work?

Instead of "Proposal: Stop providing binary distributions" maybe your
proposal is:

"Proposal: Group X will stop providing binary distributions, and instead
much more strongly encourage and advertise the work of groups
Y, Z, etc... to provide binary distributions" ?

This is all about just acknowledging limits and spreading the
workload more broadly, and also doing a better job at encouraging
users toward the fantastic work of
https://github.com/3-manifolds/Sage_macOS/releases,
the conda project, gentoo, etc...

> On Saturday, January 8, 2022 at 5:41:58 PM UTC-5 emanuel.c...@gmail.com wrote:
>>
>> Mixed feelings... See my comments on the ticket.
>>
>> Le samedi 8 janvier 2022 à 20:10:15 UTC+1, Matthias Koeppe a écrit :
>>>
>>> Comments/review please in https://trac.sagemath.org/ticket/33131
>>>

> --
> 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/0e002507-d296-49c5-a726-ef1c22b8627fn%40googlegroups.com.



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

[Dima Pasechnik]

On Sun, 9 Jan 2022, 20:46 Murray Eisenberg, <murraye...@gmail.com> wrote:

As strictly an end-user of Sage, and only on macOS, I find the prospect of not being able to download a binary (not even from https://github.com/3-manifolds/Sage_macOS/releases)

the latter is here to stay, and is already pointed to by

https://www.sagemath.org/download-mac.html

Indeed, the post was slightly mistitled.

Mac binaries mentioned there are different, mostly hopelessly broken, violating current macOS standards, etc.,  binaries.

repellent. And providing, only a Docker image instead, equally repellent. The probability is large that I would abandon Sage if binaries were not available.

On Saturday, January 8, 2022 at 5:41:58 PM UTC-5 emanuel.c...@gmail.com wrote:

Mixed feelings... See my comments on the ticket.

Le samedi 8 janvier 2022 à 20:10:15 UTC+1, Matthias Koeppe a écrit :

Comments/review please in https://trac.sagemath.org/ticket/33131

[Matthias Koeppe]

On Sunday, January 9, 2022 at 1:02:27 PM UTC-8 wst...@gmail.com wrote:

Matthias, correct me if I'm wrong, but are you proposing that we just a

better job at pointing users to binary distributions of Sage such as
https://github.com/3-manifolds/Sage_macOS/releases, conda-forge, and
of course any packaging of Sage for Linux distros, etc., instead of
pointing them at binaries that don't work?

That's right, that's the proposal, see ticket https://trac.sagemath.org/ticket/33131 (needs review).

Instead of "Proposal: Stop providing binary distributions" maybe your
proposal is:

"Proposal: Group X will stop providing binary distributions, and instead
much more strongly encourage and advertise the work of groups
Y, Z, etc... to provide binary distributions" ?

where Group X =  the upstream project SageMath.

[Dima Pasechnik]

I suppose it should not include (binary) wheels (still very much WIP) and docker or other VM-like images, right?

--
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/f02c840b-adfc-4740-8966-bf71d4474da9n%40googlegroups.com.

[Matthias Koeppe]

On Monday, January 10, 2022 at 12:11:19 AM UTC-8 Dima Pasechnik wrote:

On Mon, 10 Jan 2022, 01:31 Matthias Koeppe, <matthia...@gmail.com> wrote:

On Sunday, January 9, 2022 at 1:02:27 PM UTC-8 wst...@gmail.com wrote:

Instead of "Proposal: Stop providing binary distributions" maybe your

proposal is:

"Proposal: Group X will stop providing binary distributions, and instead
much more strongly encourage and advertise the work of groups
Y, Z, etc... to provide binary distributions" ?

where Group X =  the upstream project SageMath.

I suppose it should not include (binary) wheels (still very much WIP) and docker or other VM-like images, right?

Agreed. See ticket (needs review).