On Tue 11/02/2020 16:21, Kurt Mosiejczuk wrote:
> On Fri, Feb 07, 2020 at 09:36:30PM +0100, Björn Ketelaars wrote:
> > Diff below updates jupyter-notebook to 6.0.3. Changelog can be found at
> > https://jupyter-notebook.readthedocs.io/en/stable/changelog.html.
> > Noticeable change is that upstream dropped support for python 2.
>
> > Changes to the port:
> > - Drop FLAVOR and set MODPY_VERSION
> > - Update dependencies
> > - Drop post-install bits as it is no longer necessary to add a suffix to
> > jupyter-notebook and its tools
> > - Set @conflict- and @pkgpath-marker in PLIST
>
> > Testing:
> > - 'make test' runs successfully
> > - Run tested on amd64 in an existing jupyter-notebook environment
>
> > Comments/OK?
>
> Patch looks good to me. Will folks notice the python2 version going
> away? I suppose they might since the files going into bin/ are losing
> their -3 suffix.
>
> If we don't think that will cause any problems, ok kmos
Upgrading jupyter-notebook to this newer version is transparent, which
could cause two problems:
- Users of the python2 flavor will be presented with a python3
environment, which could cause their workbooks to 'fail';
- Users of the python3 flavor python will wonder why
'jupyter-notebook-3' isn't working any more, and will notice that
'jupyter-notebook' does work.
Although I no not think the latter is bad, we could help users by
informing them about this update. My proposal would be to add something
to current.html.
The first problem is kind of a bad one as some notebooks will need to be
converted to pyhton3 code. Users will have to invest time and effort to
update their code. Least thing I should do is inform them of this. Also
here I propose to inform users via current.html.
Is this update important enough to possibly break people's work?
Probably not. Reason for wanting to move forward is that it does not
make sense to maintain ports of tools that are not fully supported by
their upstream. In the case of jupyter-notebook support for python2 has
been dropped.
Not moving past this small hurdle will, sooner or later, cause an issue
somewhere (else). My proposal is to do it now.
Additional comments/OK's to commit this update? Opinions on the above?