SML/NJ on LPIA (i386 variant) architecture? (quick reply requested due to ML Workshop on Sunday)

View: New views
3 Messages — Rating Filter:   Alert me  

SML/NJ on LPIA (i386 variant) architecture? (quick reply requested due to ML Workshop on Sunday)

by Joe Wells :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

[ A quick reply is requested because I'd like to use my own portable
  computer to give the type error slicing demo at the ML Workshop on
  Sunday! ]

I'm trying to use SML/NJ on LPIA (Low Power Intel Architecture)
machines.  This is the architecture of the Intel Atom CPU used in many
very small portable computers.  I'm using a Dell Inspiron Mini 10
which uses this architecture.  My OS is Dell's distribution of Ubuntu
Hardy (8.10) built for “lpia”.

Supposedly i386 code runs unchanged on this architecture and the only
advantage of compiling specifically for this architecture is better
performance (mainly this means using less power).  (I don't know if
code compiled for LPIA runs on arbitrary i386 machines or if there are
additional instructions.)

Unfortunately SML/NJ behaves strangely on this architecture.  It runs,
but I've seen different behavior:

1. Pressing Control-C (which sends SIGINT to SML/NJ) while running
   code from the top-level REPL does not interrupt it.  At times
   (which I do not precisely remember) pressing Control-C has simply
   killed SML/NJ.

2. Some of our code produces different results.  I haven't had time to
   debug precisely where the difference in behavior is.  The code in
   question uses timers to control how much time is spent on a
   computation, so signals may be involved in the problem.

I've seen this problem with these versions of SML/NJ:

A. SML/NJ 110.65 built for the i386 architecture in Ubuntu Hardy.
   (This is not from Dell's Ubuntu Hardy, which does not provide a
   “smlnj” package.  Instead I downloaded the .deb files and installed
   them by hand.)

B. SML/NJ 110.70 built from source on the machine I am using.

Can anyone help?

Is there some way to run a test suite for SML/NJ that might help to
isolate the problem?

Is anyone successfully using SML/NJ on LPIA machines?

Thanks for any help you can provide!

--
With my best regards,

Joe


--
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Smlnj-list mailing list
Smlnj-list@...
https://lists.sourceforge.net/lists/listinfo/smlnj-list

Re: SML/NJ on LPIA (i386 variant) architecture? (quick reply requested due to ML Workshop on Sunday)

by Joe Wells :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Joe Wells <jbw@...> writes:

> 1. Pressing Control-C (which sends SIGINT to SML/NJ) while running
>    code from the top-level REPL does not interrupt it.  At times
>    (which I do not precisely remember) pressing Control-C has simply
>    killed SML/NJ.

Some further information.  This problem with Control-C seems to only
happen while running some of our code.  But it works on other machines
while running our code, so there is still something quite strange
happening.

--
Joe


--
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Smlnj-list mailing list
Smlnj-list@...
https://lists.sourceforge.net/lists/listinfo/smlnj-list

Re: SML/NJ on LPIA (i386 variant) architecture? (quick reply requested due to ML Workshop on Sunday)

by Joe Wells :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Joe Wells <jbw@...> writes:

> Unfortunately SML/NJ behaves strangely on this architecture.
 [ ... ]
> Can anyone help?

Okay, I've figured things mostly out.  Explanation to come later, busy
preparing for ML workshop at the moment.

--
Joe


--
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Smlnj-list mailing list
Smlnj-list@...
https://lists.sourceforge.net/lists/listinfo/smlnj-list