Subject: |
Re: IP support |
From: |
[email protected] |
Date: |
Fri, 13 Dec 1996 17:47:54 -0500 (EST) |
Jim,
On Fri, 13 Dec 1996 07:58:20 -0600 "Jim B. Kowalkowski" <[email protected]>
wrote:
> I see three basic IP developments in process, maybe four:
> 1) telescope community (Andrew/Peregrine)
> 2) IRM developments and other (Matthias/DESY/Fermilab)
> 3) HiDEOS
4) Kay-Uwe Kasemir's driver, which is PC-aware.
> My main question is this: What is the main reason that you (supporters and
> users of 1 and 2) have dismissed HiDEOS? Much (not all) of the development
> in 1 and 2 has been done after the HiDEOS development. I'm really not trying
> to sell HiDEOS or anything, I just want feedback.
I'll try and list as many reasons as I can think of.
1. The HiDEOS IP software will only work on an mv162 CPU, or (with some
changes) on a carrier board based around the IPIC chip. I had a long look
at the source-code for the HiDEOS driver before concluding that we
couldn't use it. The GreenSpring VIPC-310 and -610 IP carriers which I
had to support are *totally* dumb; you can only change the interrupt
levels by blowing a new PAL, and the base addresses are set by jumpers.
Neither setting can be discovered by reading a register, because there are
no registers. Autoconfiguration of the carrier boards is thus impossible
-- if you don't have an IP module installed you can't tell whether there's
a carrier board there or not. I forget if your software provide a means
of manual configuration or not, but you certainly need it to support the
GreenSpring carriers. The carrier driver software would also have to
provide a way of telling a module driver "sorry, you can't do that with
this carrier" and I couldn't see how with your current interface.
2. We needed to be able to install multiple carrier boards simultaneously,
and we might have needed to be able to add a GreenSpring card to a system
that was already using the mv162, hence loading two different carrier
drivers.
3. Yes, C++. I agree it's the way things are moving, but we would have
had to jump through some hoops to get Gemini (the customer) to agree to
our using it. I also failed to build the g++ cross-compiler, which didn't
do much for my confidence in the approach.
4. We found it difficult to justify loading what seemed a non-trivial
amount of HiDEOS software on top of EPICS just to be able to communicate
to one type of IP module (a similar reason for our rejection of CDI BTW).
> I think it would be nice if we could have a unified IP support package
I think it is crazy *not* to have a unified IP support package, but IMO
there wasn't anything suitable before I wrote drvIpac. I'm willing to
modify my CANbus driver to run under anything that we decide upon as a
standard, provided it supports the GreenSpring boards sensibly. I suspect
that drvIpac is the most general-purpose of all the IP support packages,
but I'm not saying we *have* to use it.
- Andrew
___
.' `. Andrew Johnson, Head of Electronics
/ Royal ) Royal Greenwich Observatory
\ Greenwich Madingley Road, Cambridge, CB3 0EZ
| Observatory Tel: +44 1223 374823 Fax: 374700
+---------- WWW: http://www.ast.cam.ac.uk/~anj
- Navigate by Date:
- Prev:
Re: "EPICS for Idiots" Steve Lewis
- Next:
Re: VME: Cannot find record type dfan Marian Zurek
- Index:
1994
1995
<1996>
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
Re: IP support Peregrine McGehee
- Next:
IP conversation, my 2 cents RON
- Index:
1994
1995
<1996>
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|