On Fri, Oct 9, 2015 at 3:46 PM, Nathan Andelin <nandelin@xxxxxxxxx> wrote:
I can't figure out what the business value IBM is getting by taking on
the maintenance / support / documentation / distribution burden of
custom IBM i forks of popular OSS projects, versus an alternative of
modifying PASE to simply compile and run them as-is (after configure,
make, make install)?
The business value began with the success of Apache based HTTP server for
IBM i. It got another jump-start with their partnership with Zend. One
thing leads to another.
Well, the partnership with Zend is definitely a different beast than
anything in 5733-OPS (in that having a partner is very different than
not having a partner). Also, one might argue that Apache was the best
option for a "turnkey" entry into mainstream HTTP hosting, and without
HTTP hosting, that's very limiting indeed. They may have deemed that
critical, in a similar way that being able to run Java on your
platform is almost not negotiable.
Lately IBM i has broaden its appeal to a new type of "developer", as
opposed to trying to serve a largely stagnant and shrinking enclave of cult
RPG III programmers.
Buck didn't say anything that goes against this. Buck is saying IBM
should serve that new type of developer in a different way than they
appear to be doing now. Specifically, I believe he is contending that
IBM should just fix PASE so that it's sufficient for whatever those
new developers need. That is, instead of having their own Python (for
example), they should upgrade PASE to the point that anyone can just
download Python's official source code from python.org and compile it
as-is in PASE. If a new version of Python comes out, then whoever
wants it can download the new Python and build it themselves, with no
help or involvement from IBM.
This would arguably be both cheaper and more effective than having
IBM-specific and IBM-maintained forks of everything. (I am not ready
to argue this myself, but Buck seems to be.)