On Fri, Jan 18, 2019 at 5:48 AM Schachner, Joseph
<Joseph.Schachner at teledyne.com> wrote:
> I'd like to add one more thing to your list of what companies will have to consider:
> 6) The ability to hire and retain employees who will be happy to program in an obsolete version of Python. A version about which new books will probably not be written. A version which new packages will not support. A version which most other companies will no longer be using, so programming only in Python 2 will place the employee at a disadvantage compared to others who have gained experience with Python 3 if they ever have to change employers.
The costs I described were _alternatives_, so what you've really
described here is one component of the costs of status quo,
maintaining a local fork, or using commercial Py2 support (options 1,
4, and 5). It is definitely a cost, though, and may at some point
become the impetus for a company to migrate to Py3... but frankly, I
doubt it. "Legacy code" is a pretty big thing in a lot of places. Just
search thedailywtf.com for "legacy" and you'll find plenty of stories
of terrible codebases held together by duct tape and blue tack, run
through a hodge-podge of different language interpreters, or even just
one interpreter but the wrong one...
By comparison to THAT, a "legacy system" that runs a clean Python 2.7
is going to be pretty tame.
Until someone says, oh hey, we should use function annotations, and
then starts running all their code (in production) via 2to3... naw,
that'll never happen, right? Oh.