[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-developer] cannot build axiom

From: root
Subject: Re: [Axiom-developer] cannot build axiom
Date: Thu, 29 May 2008 04:03:20 -0400

>>>So what will you tell them that you cannot figure out by 
>>>a proper
>>>configure script automatically?
>>I do not know how to automatically and generically figure 
>>out the
>>correct string, e.g. "debian", "macosxppc", "fedora9" 
>>,etc. It seems
>>that this would take an expert in each system to write 
>>code to probe the configuration for each system.
>I certainly don't know anything about OSes. But it seems 
>to me that there is a dumb way. If each supported OS 
>stanza includes the expert system-specific code to verify 
>(not probing to find what the OS is, but assuming the OS 
>is the one in the stanza, verify it -- only a yes or no 
>answer) that the OS is the right one, then by a process of 
>elimination, a loop going through all the OSes stanzas 
>until a match is found(or not found) should solve the 

It takes approximately 15 seconds to read the output and type the
required export lines.

To learn enough about all the systems to solve the problem, to
automate this, test it on all the systems, and maintain it in the
future will probably consume about a week of my time now, with untold
further commitment in the future.  I have no doubt it could be make to

Axiom has lived on a dozen different lisps and a dozen different
hardware and operating system combinations, for most of which I was
the primary person responsible. It probably will face a few dozen more
in the next 30 years.  I'm quite confident I can make it run anywhere
and make the build system do whatever is needed.

Given a choice between spending a week on that task and spending a
week on the thousand other tasks in the queue, I think I'll wait until
someone who considers it important enough to fix, posts a patch. You
can program. Give it a try. You know as much as I do about the subject
and everything else is available on Google. Better yet, assign the
task to an undergraduate as a class project.

It is time to stop playing with porting and language issues.

I'd rather that you choose a domain and document it so the
undergraduate can pick it up, read it, and learn how and why it works
and can learn enough to maintain it. That's something we can do that
an undergraduate cannot.

I think it is more important to document, validate, verify, prove, and
extend the system. All of those tasks have very steep learning curves
and require our expertise.

We seem to spend a lot of time focusing on the non-essential but
immediate, which is easy. We seem to be ignoring the essential and
long term, which is hard. The time has come to focus on the long term,
hard tasks.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]