Experimental Physics and
Industrial Control System

Pete Jemian <[email protected]> · Sat, 08 Jun 2013 07:45:43 -0500

Emmanuel:
On 6/7/2013 5:22 PM, Emmanuel Mayssat wrote:

I am looking for a solution where just changing the PVname AUTOMATICALLY
changes the autosave configuration.

You might be able to construct some shell script to compose a .req file

in the same directory as the .db file, then create a Makefile rule to

run this script when the .db file is changed. A "make" would be

required to implement the automation. Perhaps this last step could be

configured into a specific and capable editor used to change the .db

file? Then the process would be automated when the .db file is saved

from the editor.

It is likely you have a design or use case that is different than what

we see. Indulge me a deeper explanation of our use case with regards to

.req files.

We avoid doing automatic generation of .req files since we consider the

items for autosave are only those required to be restored upon restart

of the IOC. (This might be where our use cases differ.) The choice of

what goes in the .req file associated with a particular .db always

requires some human judgment.

in your case, if I understand correctly, changing a PVname ALSO requires
a change in the req template or macros.

This is correct.

Obviously, compared to dbgrep or autosaveFields, the advantage of the
req template for developers is that it is easy to identify what is being
saved and what isn't. But to date, most of the problems I have with
autosave is due to PV name changes and out-of-sync req file.

And that raises the real question for me here.
What part of the PV name changes for you and why was this required?

At APS, we tend to think of PV names as being constructed from parts.

The first part describes the IOC or system, almost always using the P

macro. The next part or parts might describe a particular instance of a

database on that IOC or system. In the motor example I sent, this is

the M macro. The final part describes the record(s) in the particular

.db file.

The .req file is then (re)constructed by the database designer (a human)

based on knowledge of the features provided by the .db and a decision of

what needs to be restored on IOC restart. The macros used in the .req

file are almost always the same as those used for the .db file to

simplify things. In many instances of .db files, there are records and

fields defined that should not be added to the .req file. For instance,

the synApps motor.db file, it makes no sense to add the

$(P)$(M)_ableput, $(P)$(M)_vCh, and $(P)$(M)_twCh records to autosave.

https://subversion.xray.aps.anl.gov/synApps/motor/trunk/motorApp/Db/motor.db

https://subversion.xray.aps.anl.gov/synApps/motor/trunk/motorApp/Db/motor_positions.req

https://subversion.xray.aps.anl.gov/synApps/motor/trunk/motorApp/Db/motor_settings.req

It helps for the .db and .req files to be co-located in the

<top>/someApp/Db directory. This helps on several levels: first, we

have infrastructure that automatically looks on this path, and more

importantly, it supports the discipline we must have to consider changes

in both .db and .req files together. Several IOCs can use the same pair

of .db and .req files, changing only the values of the macro parameters.

 Additional macro parameters might be used to configure specifics

within the .db file such as default BAUD of a serial port.

Changes in the .db file record names often have consequences on

downstream software such as autosave, user interfaces, PV logging, and

processing. We make efforts to minimize such changes.

Regards,
   Pete

--
----------------------------------------------------------
Pete R. Jemian, Ph.D.                <[email protected]>
Beam line Controls and Data Acquisition, Group Leader
Advanced Photon Source,   Argonne National Laboratory
Argonne, IL  60439                   630 - 252 - 3189
-----------------------------------------------------------
   Education is the one thing for which people
      are willing to pay yet not receive.
-----------------------------------------------------------

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System