Re: Sequencer/SNL and PVA?

["Johnson, Andrew N. via Tech-talk" <tech-talk at aps.anl.gov>]

Hi Ben,

On Oct 6, 2020, at 4:24 AM, Ben Franksen via Tech-talk <tech-talk at aps.anl.gov> wrote:

Am 06.10.20 um 10:25 schrieb Ralph Lange via Tech-talk:

Since we started developing a number of "pure" pvAccess servers,
users/developers have been asking how to do state machines that connect to
them.

Is there any experience or plan to add support for pvAccess to the
SNL/Sequencer?

I have no concrete plans. There are basically two reasons:

(1) I can't see how to reconcile the SNL data model with PVA. In CA you
have the basic primitive types which are mapped to the basic SNL types;
or arrays of those which are similarly mapped to SNL arrays; and then a
fixed set of meta data on top of that. PVA provides a richer set of data
types that does not map easily to a language that is, give or take, a
variant of C. Also, meta data is not as uniform in PVA as in CA. Even if
support is restricted to Normative Types, I am not sure we can map all
of them to SNL types in a reasonably useful way.

So there are two kinds of mappings between the data models, and it looks like you’re wondering how the sequencer could provide broader access to the full PVA data model. However there is a simpler approach which is for the sequencer to just support a limited subset of the PVA data model that it can use without changing the sequencer at all, in a similar manner to how QSRV's pva link type supports the simpler IOC data model over PVA.

I just searched tech-talk and found the thread from 2010 where you asked the community about dropping support for the pv layer. From that thread it looks like you decided to remove the pv support completely, although I’m still seeing files for it in the src/pv directory in the 2.2 versions at least, was it the 2.3 branch where you removed it? Is it still functional in 2.2, or might it be feasible to resurrect it or add something like it to implement an interface to PVA?

One possibility that just occurred to me (although I realize it’s probably totally impractical) would be to see if the sequencer could interface to the link-support APIs that the IOC now uses for JSON link types. That would give us a PVA implementation with almost no PVA-specific work, although as I said I’m not sure how much effort it would take to make that work.

- Andrew

-- 

Complexity comes for free, simplicity you have to work for.