Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Wed, 3 Dec 2008 15:38:49 -0600

On Wednesday 03 December 2008 14:32:09 David Dudley wrote:
> Is there some specific reason that some of these other records *aren't* a
> part of base?

> For instance, Archive seems pretty handy from reading the documentation, as
> does Timer, and especially ePID (which isn't in the wiki).

AFAIK the archive record is pretty specific to the DESY archiver, which is 
different to the channel archiver.  The timer record is also included in the 
unbundled Mizar support and probably rather specific.  There used to be a PID 
record in Base, but it had a few problems and various other people developed 
enhanced versions, of which I believe the ePID to be the best.  However 
either they were never offered back to Base or it was felt they were better 
maintained independently.

> How did the specific records in base get -canonized- to be a part of base?

History and authorship/maintenance.  The majority of the record types in Base 
date back to the original GTACS code.  A number of record types have been 
added and removed over the years; in between 3.13 and 3.14 we removed several 
as part of the un-bundling of all hardware-specific device and record support 
from Base.  That was when the Mizar support module got created for example.  
I added the aSub record in R3.14.10 because there were many people using the 
genSub record from which it was derived, and it's a very useful generic 
thing.

> Is there any reason for these other records to *not* be considered for
> inclusion into base?

They are better off being separately maintained; this spreads the workload 
across the community, and means that if nobody uses them they naturally stop 
taking up maintenance effort.  As it is there are a few record types in Base 
that are probably not being used, so they'll probably get unbundled for the 
next major release.

My aim is actually to reduce the size of Base rather than increase it; one of 
my plans is to split Base into 3 or more parts - Core, CA (or CAC and CAS) 
and IOC and allow them all to evolve separately.  I'm keen to get other 
people to create different EPICS Distros containing a particular collection 
of modules that make up a complete EPICS toolkit.  I expect synApps to become 
one such Distribution.

HTH,

- Andrew
-- 
Talk is cheap. Show me the code. -- Linus Torvalds

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System