HP3000-L Archives

April 2002, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
James Clark Jr <[log in to unmask]>
Reply To:
James Clark Jr <[log in to unmask]>
Date:
Mon, 15 Apr 2002 13:21:12 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (58 lines)
I am sorry if I led you do believe it needed to be supported. It should only
be supported on the given OS, that being MPE/iX or emulator. But as for
support on other platforms, it would be a waste of time in my estimation.
Just get your code running on IEEE then move it over to some other product.
This would be hard for those who are still running on an old HP Classic
machine, but you should be able to pick up an old PA-RISC machine for a song
and do your conversions there. The smallest PA-RISC machine blew the doors
off the series 70, given enough memory that is.
So don't take me wrong, I was just mentioning concerns for those moving from
one to the other. Because the fastest load and unload of a DB is native and
image would not care what data actually had in it, thus after changing the
DB to IEEE reals and reloading all your reals would be trashed except for 0.

        James

-----Original Message-----
From: [log in to unmask] [mailto:[log in to unmask]]
Sent: Monday, April 15, 2002 1:06 PM
To: James Clark Jr
Cc: [log in to unmask]
Subject: Re: [HP3000-L] From Pascal/V to PASCAL/XL


Hello James.

Thank you for your feedback.

I agree that binary migration (of the application) of CM applications
or programs interacting with CM applications could be a problem. As
far
as i understand, a native application usually should not make use of
the HP3000 floating point format.
If necessary we could convert between HP3000 and IEEE floating point
format on the fly (for a specific application).

A binary migration of data could simply convert the values
transparently
similar to what we do currently through text mode.

On precision: AFAIR the R2 variable has an additional bit for the
exponent vs. IEEE float but lacks one in the mantissa.
The R4 has three bits more in the mantissa than the IEEE double.
If this makes a difference in real life i don't know.

This comes down to: Does it really make a difference to support the
old HP3000 floating point format natively? I would guess it will
always be alien (read troublesome) on all platforms besides the
HP3000.


Michael


<snip>

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2