Experimental Physics and
Industrial Control System

"Kasemir, Kay via Tech-talk" <tech-talk at aps.anl.gov> · Thu, 29 Jul 2021 13:18:50 +0000

> I guess if we restrict to only generating the .req file, this way the solution will be quite nifty. Adding the xml generation library to it will make it bulky.
> I suggest a two stage solution, the makeArchiveConfigFile will make a .req file. Each line of this .req file will contain all the necessary archive information per PV. This file can subsequently be parsed using a python app for generating a CSS archive compatible xml config file.
> We can extend the AutosaveFileFromDbInfo function and request the maintainer if this can be a standard feature of autosave.

The information that you need is
1) How to sample: scan or monitor
2) How often to do that, i.e. scan period or estimated update period
3) PV name, which usually is the record name with implied .VAL, but in rare cases you might for example want the .RBV

Example:

  # Request archiving on change
  info(archive, "Monitor, 00:01:00")

  # Scan
  info(archive, "Scan, 01:00:00")

  # VAL field is the default
  info(archive, "Monitor, 00:01:00, VAL")

  # In rare cases, might need to archive other fields
  info(archive, "Monitor, 00:01:00, RVAL DMOV")

Why do you want to add this to the autosave/restore code? They're not related.
Why do you want to create some arbitrary intermediate file format?
Why not create the XML right away?
Yes, I would avoid adding an XML library, but "printf" will suffice since you know what you'll write. Just check the PV names for a very limited number of potential characters to escape, which I think is basically '>' and '<'.

So something like ` makeArchiveConfigFile("/path/to/ioc_xyz_archive.xml")` at the end of st.cmd will create one file per IOC, and you can then use xinclude to merge the per-IOC files into one per archive engine.

-Kay 

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System