[Info-vax] CPython has removed OpenVMS support

[Craig A. Berry]

On 12/21/13, 5:34 PM, Simon Clubley wrote:

> It might be a interesting exercise for anyone using them to see if any
> VMS support for other third party languages and environments is in the
> process of being removed or has been proposed for removal.

What most open source projects are looking for is the opposite of
"drive-by" porting, or in other words, they frown on what's known as a
"patch bomb," a big set of changes that gets everything working with a
new feature or on a new platform, whereupon the submitter disappears and
leaves the hard work of maintenance to busy people who may be
sympathetic but have no experience with or knowledge of said feature or
platform.

Sometimes people go to extreme lengths to support what are to them
"foreign" platforms. You can see that in the Python thread cited
earlier. Someone acquired Itanium hardware and begged and pleaded for
media so he could install VMS and get to work keeping the Python port
up-to-date. He hadn't even gotten far enough to realize he would need
licenses as well as media. But no one who cares about VMS was involved
in Python development enough to steer him in the right direction.

Unfortunately drive-by porting is about the only kind DEC/Compaq/HP have
ever done, Apache being one of the more painful examples. They produced
a very nice port of a stable release and then were disappointed when the
Apache folks wouldn't accept radical changes to a maintenance branch. So
we're still stuck on 2.0.x.

But as far as your question, every port is at risk unless people are
(week in, week out, year in, year out): reporting and fixing bugs,
submitting patches (against the development branch!), running the test
suite, reporting test results to whatever continuous integration package
is in use, and generally participating in the community.

> For example, where does Perl stand in terms of VMS support ?

"Support" is a word that means different things to different people in
different contexts. As far as I know, HP no longer includes support as
part of OpenVMS support for a Perl package they distribute, and in fact
they no longer distribute a package.

As far as the open source project is concerned, VMS support is still
very much present in the code and in the test suite. There are over 2200
test scripts in the test suite, and currently only one of them is failing.

A few years ago someone proposed removing the VMS support from the code
and was quickly chased away, not so much by me but by several key
leaders in the developer community who are proud of Perl's support for
eccentric platforms.

But we do deprecate platforms when no one is showing up to do the work
of supporting them. I think Mac OS Classic finally bit the dust recently.

> BTW, is that a purely third party maintained port, or has VMS support
> been added back into the mainline Perl codebase ?

It's not an either/or question. I'm the third party, but I have a commit
bit and the port has been maintained continuously as part of the
mainline distribution for a couple of decades. According to Ohloh.net
(<http://www.ohloh.net/p/perl/contributors/19391777492879>), I've
contributed 121 changes this year and 5,160 in the thirteen years I've
been involved. That's what I mean by the opposite of drive-by porting.

It has to be said though, that the bus factor is pretty low. If I were
to stop doing this, it would take a year or three, but the Perl folks
would eventually (and I think for most of them reluctantly) excise the
VMS port if no one were actively, visibly maintaining it.