Re: [tbox-l] looking for Python 2.6 on AIX

[Bryan H. <hb...@us...>]

Reza Arbab <ar...@au...> wrote on 06/16/2009 09:35:16 AM:

> On 15 Jun 2009, 04:16PM CDT, Jørn Amundsen <Jor...@nt...> 
wrote:
 
> >             It is difficult to understand why IBM can not cooperate
> > with one of its long term University customers, for instance by
> > donating a system, to a U, provided the U agrees to administer the
> > system and keep it on to the Internet for X years. Then an open source
> > project could be set up on the system, where IBM individuals and other
> > package builders could contribute to the community.
> 
> UCLA used to do this.  I don't know if there was an arrangement with 
donated
> machines, though. 

I'm sure availability of computing systems is not a significant factor in 
whether an operation like this can exist.  The primary factor is packaging 
and porting labor.  I don't think IBM would have any trouble contributing 
that either, but for some parts of IBM, there is the liability issue. 
Others, including universities and users of AIX don't feel the liability 
issue, but apparently can't scrape together the labor on a longterm basis.

> It's funny, I think in the beginning this Toolbox may have been intended 
to 
> consolidate some of the other AIX porting projects.  Indeed, a good 
number 
> of users (with little knowledge of open source) were attracted because 
they 
> assumed these packages were somehow under official IBM warranty or 
support,
> which they certainly aren't.

That's not how I understand its history.  I am an AIX user, by the way, 
not a developer, and while I work for IBM, I don't work in the division 
that produces AIX.  I saw the toolbox come on the scene ca 2002 as part of 
an IBM initiative to tap the new Linux market.  It was called the AIX 
Toolbox for Linux Applications for POWER Systems and connected with a 
product branded AIX/L.  As I understand the marketing pitch, the idea was 
for people to buy AIX and use it to develop applications to be run on 
Linux systems.  (The answer to the obvious question -- why not just 
develop on Linux -- was that AIX is more reliable and has better technical 
support and bug fixing than available Linux products).  But _I_ just 
called it the "GNU upgrade for AIX," since my AIX system is far more 
serviceable with GNU ls, make, etc. instead of the ones that are part of 
AIX.

One of the seven or so CDs on which AIX was delivered was a CD containing 
the toolbox.  IBM also offered the toolbox website for newer stuff.

I don't know to what extent IBM just gathered ports from other places, but 
it was always my impression that IBM did a massive porting effort itself 
to create a complete set of tools that worked on its product.

> Now that the Toolbox has become so stale, the situation is reversing and
> 3rd-party alternatives are the attractive option.  There are a number of 

> really good ones out there right now.  Google for "aix open source".

Marketing initiatives don't last more than 5 years.  This one was 
especially tenuous; I suspect it did not generate the sales its creators 
expected.  Are there people who would use AIX instead of Linux if they 
could easily get all the current applications?

> > IBM individuals are doing huge contributions to GCC and a lot of other
> > excellent open source initiatives, so how come it is not possible to
> > establish an open source project on packaging RPM's for AIX ?
> 
> The perceived risk of distributing other people's code was enough for 
> our various lawyers and steering committees to completely cripple the 
project 
> with process and red tape.

I do know that in at least some of the areas where IBM distributes code 
written by the open source community it has the same fears and process, 
but because the value of distributing is so high in those areas, IBM 
accepts the risk and puts in the resources to go through the process.

It's hard for many of us to understand how it might cost IBM millions of 
dollars if it distributes a port of Python, but on the other hand, if 
there's no substantial upside, I can see where it might not be worth 
taking a chance.  It costs money just to evaluate the risk.


It shouldn't be a big deal.  IBM is obviously in the best position to 
offer a toolbox like this, but anyone else could do it.  User groups have 
always done stuff like this.  Companies invest a lot of money in AIX 
systems; they should be willing to pay for the open source additions to 
it.

--
Bryan Henderson                     IBM Almaden Research Center
San Jose CA                         Storage Systems