Experimental Physics and
Industrial Control System

Marty Kraimer <[email protected]> · Tue, 11 Feb 1997 16:18:41 -0600

During the APS run starting at the beginning of last december, epics
base release beta4 was used for the Linac, Par, and all of the RF
systems. The only change made to beta4 is that the sequencer problem
discussed in tech-talk was fixed.

We are currently nearing the end of a scheduled shutdown. When we
startup the booster iocs will also be running with base release 3.13.

All of our MCR workstations have been upgraded to solaris. The
extensions are all build against the beta4 release of channel access.

Since beta4 the main changes to epics base have been to base/config and
to make files. The changes were mostly the work of Kay-Uwe Kasemir, who
has been working at LANL. These changes are to support the port to win95
and winNT. In epics base, each Makefile.Unix is replaced by a
Makefile.Host. In addition a new file config/RULES.Host replaces the old
config/RULES.Unix. Epics base will continue to support RULES.Unix so
that application S/R environments, which use it, will continue to work.
Ultimately, however, all applications will have to start using
RULES.Host.

We have converted the Linac, Par, and Booster to use the beta7 release.

At APS the systems that have been converted to 3.13, are now also using
CVS instead of SCCS. This, in addition to the change to release 3.13,
required major changes in our Application Source Release control system.
The APS EPICS WWW documantation has a new document, "IOC Applications:
Building and Source/Release Control R3.13.0beta7 ",which describes the
new system.

The above document replaces chapter 2 "Building IOC Applications" of the
beta4 version of the Application Developer's Guide. Thus the utility
makeBaseApp works as part of the S/R system.

The main changes to makeBaseApp since beta4 are 1) It uses Makefile.Host
instead of Makefile.Unix, and 2) iocs are now booted from a separate
directory structure rather than from bin.

In a couple of weeks we should have a beta version ready that supports
win95 and winNT with the following qualifications:

1) The sequencer has not been thouroughly tested.
2) The logServer has not been ported (problems with signals, etc.)
3) The ca repeater (on winXXX) must be manually started.
4) With real devices the ioc code has been run on a pentium.
   It has been loaded on a 486 but not tested with I/O devices.

We still have two build issues to resolve before we can say that we can
successfully build epics base on win95 and winNT. 1) Several base shell
scripts will be rewritten in perl, and 2) The makeBaseApp utility uses
soft links in two places.

Kay has actually rewritten the scripts in perl but the new scripts have
not yet been committed to the epics base repository.

An aside about the choice of perl. The tech-talk discussion about the
choice of a scripting language seemed to revolve more around high level
applications rather than building. Especially for building epics base we
feel that perl is a better choice than python or tcl. This choice should
not influence the choice of scripting languages for high level
applications.

The APS WWW EPICS Documentation page has been simplied. Anyone who has
contributed documentation should review it and tell us if something is
incorrect or missing.

Marty Kraimer

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System