OpenMandriva Lx 4.0 Punts Python 2 Out Of Its Base OS

[schmidtbag]

I've read some git issues about this on some projects and one common reason is "I know Python 2, not 3, so I don't want to port it". In those cases we'll have to use legacy environments or hope someone who knows what they're doing picks up the torch.

With how big some projects have become, I think some are considering switching to something else like Rust or Go. Makes sense. If you're forced to port due to a language EOL, might as well pick something a little more low level with long term support so this scenario doesn't happen again anytime soon.

Personally, I interpret that as "I'm too lazy". Python 3 isn't hard. It's a little more picky about encoding, it's more consistent with its use of parenthesis, and from what I recall it's a little better about how it handles division. I've once ported a Qt-based program involving thousands of lines of code from Python 2 to 3 and I did it all in about a day. The vast majority of that time was compensating for the differences in Qt, not so much Python itself.

[berolinux]

I wonder, does sagemath work for you with python 3? I tried searching both "mandriva sagemath" and "fedora sagemath", but didn't get any links describing its dependencies. AFAIK sagemath were in process of transitioning to python3, but didn't quite finish it, that's why I'm wondering if it works purely with python3 on Mandriva and Fedora.

Starting to get there, but not fully done yet:

https://wiki.sagemath.org/Python3-compatible%20code

It's one of the reasons why we still keep the python 2 packages around.

[skeevy420]

More interesting why all these upstream projects don't migrate to Python 3. Especially since the deadline had been known for years.

I've read some git issues about this on some projects and one common reason is "I know Python 2, not 3, so I don't want to port it". In those cases we'll have to use legacy environments or hope someone who knows what they're doing picks up the torch.

With how big some projects have become, I think some are considering switching to something else like Rust or Go. Makes sense. If you're forced to port due to a language EOL, might as well pick something a little more low level with long term support so this scenario doesn't happen again anytime soon.

[Hi-Angel]

More interesting why all these upstream projects don't migrate to Python 3. Especially since the deadline had been known for years.

For sagemath I think it was lack of resources.

[S.Pam]

More interesting why all these upstream projects don't migrate to Python 3. Especially since the deadline had been known for years.

[Hi-Angel]

I wonder, does sagemath work for you with python 3? I tried searching both "mandriva sagemath" and "fedora sagemath", but didn't get any links describing its dependencies. AFAIK sagemath were in process of transitioning to python3, but didn't quite finish it, that's why I'm wondering if it works purely with python3 on Mandriva and Fedora.

[RahulSundaram]

It has started 2-3 years earlier -- actually, more. We (OpenMandriva) have been gradually working on it since our 2014 release. We only arrived at the point where we can safely remove Python 2 from our install image now.

The problem is that a number of upstream projects don't care (even now), and the big distros that have enough people to get something like this done in a reasonable time frame always seem to move slower than their much smaller counterparts.

As one of the larger distros, Fedora has dropped Python 3 off its primary image (Workstation edition) a while back. The current Fedora goal is dropping Python 2 entirely and that is of course a much bigger undertaking. There is certainly a lot of effort being spend on it and substantial progress being made which smaller distros can lean on, as well, because they are being upstreamed.

[DanL]

This should have started a good 2-3 years before the announced EOL date not less than a year.

Ummm, it has. The transition process has gone on for multiple releases for Debian, Fedora, etc. They try to make it as painless as possible for users and packagers.

What did you have in mind? Instantly ripping Python2 out and all packages that depend on it on the day you read about the 2020 EOL date? LOL. I hope you're not maintaining any software I use.

EDIT: I type too slowly and got my thunder stolen

[berolinux]

This should have started a good 2-3 years before the announced EOL date not less than a year.

It has started 2-3 years earlier -- actually, more. We (OpenMandriva) have been gradually working on it since our 2014 release. We only arrived at the point where we can safely remove Python 2 from our install image now.

The problem is that a number of upstream projects don't care (even now), and the big distros that have enough people to get something like this done in a reasonable time frame always seem to move slower than their much smaller counterparts.

[wizard69]

This should have started a good 2-3 years before the announced EOL date not less than a year.