On Thursday, October 29, 2020 10:49:27 PM WET Barry wrote:

> After the trauma of the 3.0 release it would be mad to do a python 4.0 for

> such a trivial reason as packagers that assumed, wrongly, that version can

> only have single digits.

> Barry

This now is an epistemological discussion. :-)

I would not say that the assumption was wrong, IMHO it was right.

People did not assume that version had a single digit they simply took what it worked. Because if you look back in history both python 1 and python 2 had only single digit so why should we add another level of complexity?

The same already happened before and in this case it affected python when the linux major version went from 2 to 3.

Because if we go to a defensive programming, as an example, we should always check the return value of printf (C/C++). There are places where it makes sense but in most places and in most cases that only makes the code longer, hard to read and hard to maintain.

And that was one of the reasons why there are projects changing from the semantic versioning to a scheme where only the major version number is relevant, e.g. gcc, octave, the browsers...

My original remark was on purpose ignoring that trauma, that in part was also self-inflicted, and joking about that. :-)

After all this is one way to deal with trauma, or not? :-)

Regards,

Jos� Ab�lio