| 1 |
On Thu, 2019-12-05 at 03:56 +0100, Thomas Deutschmann wrote: |
| 2 |
> Hi, |
| 3 |
> |
| 4 |
> On 2019-12-05 01:15, Aaron Bauman wrote: |
| 5 |
> > * Removal in 30 days |
| 6 |
> |
| 7 |
> Why? I understand that Py2 will reach EOL upstream status but we all |
| 8 |
> know that Py2 will *not* disappear and stop working in 26 days... |
| 9 |
> |
| 10 |
> There's no reason to mask/remove currently known working software. |
| 11 |
> |
| 12 |
|
| 13 |
Yes, there is. Not saying about any particular package out there but |
| 14 |
the Python team is *overwhelmed*. We can't reasonably be expected to |
| 15 |
maintain 1200+ packages, many of them requiring a lot of work. |
| 16 |
|
| 17 |
It's easy to claim everything is all right without actually working |
| 18 |
on it. Many of those packages may not pose problems in the next weeks. |
| 19 |
Some of them probably already take part in the big 'target mishap' when |
| 20 |
their dependencies dropped py2 support, and more will in the coming |
| 21 |
weeks, months. |
| 22 |
|
| 23 |
Now imagine that 500+ packages are depending on pytest that doesn't |
| 24 |
support py2 anymore. And that number is probably far smaller than |
| 25 |
reality because a lot of packages are bad quality and don't run tests |
| 26 |
at all. |
| 27 |
|
| 28 |
You people need to start thinking of terms of real benefit to users. |
| 29 |
Keeping old, unmaintained, semi-broken packages has little benefit to |
| 30 |
users. Quadruplicating maintenance burden effectively harms *active* |
| 31 |
packages, and that is much more painful to users. |
| 32 |
|
| 33 |
Do you think we'd be stuck with unmaintained Python 3.6 in stable if |
| 34 |
people actively kicked stuff we can't maintain? Do you think we'd be |
| 35 |
stuck with Python 3.7 packages being *unkeyworded* on almost all arches? |
| 36 |
|
| 37 |
-- |
| 38 |
Best regards, |
| 39 |
Michał Górny |
Archives