develooper Front page | perl.perl5.porters | Postings from October 2011

Re: Speak up now about your use of EBCDIC or WE WILL REMOVE IT in afuture release of Perl

Thread Previous | Thread Next
Nicholas Clark
October 1, 2011 02:11
Re: Speak up now about your use of EBCDIC or WE WILL REMOVE IT in afuture release of Perl
Message ID:
On Sat, Oct 01, 2011 at 09:49:03AM +0200, Jarkko Hietaniemi wrote:
> I don't know about "bad blood"... irritation, maybe.  Amazement at the 
> pigheadedness of IBM, why they don't see how the current model doesn't
> work.  Their current model of maintaining an obsolete fork doesn't work.
> Well, to be fair, they do every couple of years throw a bunch of poorly
> documented patches over the wall.  But they have never responded to any
> of our feedback and/or questions about the patches.  Some of them have
> been integrated, some haven't.
> Personally, I'm mightily irritated at my and NI-S' work being wasted
> this way.  And I think we are fooling ourselves if we claim that it'll
> be easy to "merge the changes back in".  It was a mess.
> The assumptions of ASCII run deep.

I agree with everything you write. Particularly frustrating is the lack of
responses/feedback to *our* questions about their patches.

I'll add something (which you may not agree with). I have no confidence in
the quality of the work that IBM did. Whatsoever. The employees assigned to
the porting task seemed to have no idea of the implications of what an
EBCDIC platform meant

[can't find the thread where this was, but they were confused as to why
the output was garbled if they told perl on an EBCDIC machine to convert
output to "ISO-8859-1", but everything was readable by default. They didn't
seem to grasp that

* the default output of a perl on EBCDIC system was in EBCDIC
* that between them and their desktop there would have to be conversion of
  octets, where the octet mapping is always for EBCDIC->ASCII
* that for these reason, they would not be getting the same output as
  when they ran the self-same programs on their desktop, or a Unix system

In this thread, there is a revealing honest answer:

Q: Do you have any idea *why* this change makes things work?
A: I havent been able to figure out why.

My firm belief is that most of the changes were "hack it until the tests
pass", with no real understanding of what they were doing or why.

That does not strike me as a way to produce maintainable code.

In contrast, IBM's third line EBCDIC support, based in China, seem to
know exactly what they are doing, and have sent good clear patches which
we've been happy to apply. We'd love them to send us more.

Nicholas Clark

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About