Re: StreamDevice for an odd float format

[Ralph Lange via Tech-talk <tech-talk@aps.anl.gov>]

Please have a look at the StreamDevice talk that Dirk Zimoch gave yesterday at the EPICS Meeting:

https://indico.cern.ch/event/766611/contributions/3438399/attachments/1856989/3050490/News_from_StreamDevice.pptx

It looks like rewriting input using Regular Expressions is now supported, which might help in this case.

Cheers,
~Ralph

On Thu, 6 Jun 2019 at 02:00, Brown, Garth via Tech-talk <tech-talk@aps.anl.gov> wrote:

Thanks, I like your approach. I'll give it a try.

Garth

---

From: Priller, John <priller@frib.msu.edu>
Sent: Wednesday, June 5, 2019 2:05:30 PM
To: Davis, Mark; tech-talk@aps.anl.gov; Brown, Garth
Subject: Re: StreamDevice for an odd float format

 

Z is just the letter I selected when I wrote my protocol converter add-on.  There's a couple issues doing it this way but it seemed the best way forward for us at the time.

One, the interface to Stream is unofficial and can (and does) change from version to version.

Two, several necessary header files aren't exported as part of the package.  We have a lightly-patched version of Stream we link against that does export those files

---

From: tech-talk-bounces@aps.anl.gov <tech-talk-bounces@aps.anl.gov> on behalf of Davis, Mark via Tech-talk <tech-talk@aps.anl.gov>
Sent: Wednesday, June 5, 2019 4:01 PM
To: Brown, Garth; tech-talk@aps.anl.gov
Subject: Re: StreamDevice for an odd float format

 

You can add your own format converters to StreamDevice.

John Priller (prioller@frib.msu.edu) at our lab has written the code for several odd format like this (see attached code).  I believe the one you want is the %ZHF one.

Mark

Mark Davis NSCL/FRIB Control Systems Software Engineer davism50@msu.edu

On 6/5/2019 1:58 PM, Brown, Garth via Tech-talk wrote:

Hi All,

I'm trying to use StreamDevice to talk to a device that is almost, but not quite, giving me data that the format converters (as I understand them) can convert.

It’s an ASCII representation of the IEEE 32 bit floating point format, e.g. the raw data for a value of 28.0 is

0x34 0x31 0x45 0x30 0x30 0x30 0x30 0x30

Which is the ASCII code for

41E00000

Which is the IEEE-754 floating point representation of 28.0.

I think that if the raw data were 0x41E00000, %R would be able to convert it. And if the raw data were

0x32 0x38 0x2E 0x30 (ASCII for 28.0)

then %f would be able to convert it.

Before I start writing calc records to convert the 32 bits to a float value, I figured I'd ask if someone had a less brute-force approach. Maybe a StreamDevice format converter trick I don't know about?

Thanks,

Garth