> On Aug 24, 2016, at 12:59 PM, Dima Spivak <[email protected]> wrote:
>
> While I definitely appreciate your position, Allen, I agree with Andrew
> here. As much as I love Bash (just ask my frustrated colleagues...),
It's funny that you brought up bash in this context.
There are lots and lots and lots of times I wish test-patch would have
been built with bash 4. It would make life a lot easier in lots of places. So
why don't we use it? Because it would cut out a chunk of the user base.
I view the python 2.7 argument (at least today) in the same vein. I'm
not convinced that we're in a place where saying that 2.7 is the minimum is
viable, especially when parts of the ASF's own infrastructure can't run it.
That's extremely hypocritical.
> I worry that we'll close the door on a lot of useful
> contributions if we maintain this as a requirement.
I think it should be understood by contributors that one of the
challenges of working with QA tools is that there is an expectation that they
should work in particularly rugged environments. They generally don't get the
privilege of dictating requirements. These tools are being used to help make
some other project better. Apache Yetus is not the core of the end product,
just one of many cogs. The more requirements we put in place, the fewer places
the project will be invited to help out.
