1 |
On Mon, May 11, 2020 at 10:02 PM Raphael MD <raphaxx@×××××.com> wrote: |
2 |
> |
3 |
> now suffering with python 3.6/3.8 dependency mess. |
4 |
> |
5 |
|
6 |
On anything but a simple system it seems very difficult to deal with |
7 |
the python update without overriding the python flags on at least some |
8 |
packages. |
9 |
|
10 |
I recommend maintaining these in a separate config file to make it |
11 |
easier to clean these changes up eventually. |
12 |
|
13 |
Following the news item and enabling 3.6 and 3.7 in the interim is |
14 |
probably your best course of action to minimize the amount of |
15 |
micromangement. |
16 |
|
17 |
You shouldn't see much asking for 3.8 just yet, at least not when |
18 |
using stable keywords. If you're using testing keywords, well, thanks |
19 |
for testing, and you can see why they call it that. |
20 |
|
21 |
In a few weeks I think it will settle down, and it isn't quite as bad |
22 |
if you're updating daily/weekly since the number of changes is more |
23 |
moderate. |
24 |
|
25 |
Much of the problem stems from the fact that we're trying to rapidly |
26 |
get up to date on python. That means that various packages are in |
27 |
various states of support for various versions of python. Also, |
28 |
default settings for building python were changed before all the |
29 |
packages supporting them were stabilized. I can't really speak for |
30 |
those doing that but I suspect it is in part to avoid having to wait |
31 |
for bugs for all the impacted packages to get resolved, or to obtain |
32 |
permission to force them to be removed. |
33 |
|
34 |
-- |
35 |
Rich |