Experimental Physics and
Industrial Control System

Jeong Han Lee via Tech-talk <tech-talk at aps.anl.gov> · Mon, 14 Dec 2020 20:36:48 -0800

Hi Felix,

The similar stories ended with no community solution, however, please
look at the following threads :
https://epics.anl.gov/tech-talk/2012/msg02173.php
You may realize the better idea while reading these threads. But I
like your first option, which is a simple approach.

HTH,
Han

On Mon, Dec 14, 2020 at 12:53 PM Matt Rippa via Tech-talk
<tech-talk at aps.anl.gov> wrote:
>
> Hi Felix,
>
> Some solutions to package applications have been discussed here (in the debian context): https://epics.anl.gov/tech-talk/2018/msg02078.php
> I agree with you, it would be great to have some generic rules in EPICS base supporting both rpm and debian packages. I suspect the differences
> are great enough this becomes very messy.
>
> -Matt
>
> On Mon, Dec 14, 2020 at 6:23 AM Felix Krämer via Tech-talk <tech-talk at aps.anl.gov> wrote:
>>
>> Hi all,
>>
>> Currently, I do some contractor work for Gemini for a new Application
>> Development Environment based on RPM.
>>
>> In this context, I am working on one part of it to provide a framework
>> for softIOC installation, everything readily set up with conserver
>> configuration, procServ etc. after installing a respective softIOC RPM.
>> I do this development work based on a makeBaseApp.pl generated softIOC
>> example from the EPICS Application Developer's Guide (
>> https://epics.anl.gov/base/R3-15/6-docs/AppDevGuide/GettingStarted.html#x3-60002.2
>> ). For installing EPICS software using RPM we have to _relocate_
>> installed binaries, scripts etc. This works well, principally, but
>> regarding the example softIOC there do arise a few questions:
>>
>> (1) The startup file under iocBoot (st.cmd) is not installed anywhere.
>> If INSTALL_LOCATION is not redefined, then this is not a problem, but if
>> I want to keep the sources distinct from the product, this seems not to
>> be consequent. I changed the Makefile from
>>
>> TOP = ../..
>> include $(TOP)/configure/CONFIG
>> ARCH = $(EPICS_HOST_ARCH)
>> TARGETS = envPaths
>> include $(TOP)/configure/RULES.ioc
>>
>>
>> to
>>
>> TOP = ../..
>> include $(TOP)/configure/CONFIG
>> ARCH = $(EPICS_HOST_ARCH)
>> TARGETS = envPaths
>> SCRIPTS_IOC_Linux += st.cmd
>> include $(TOP)/configure/RULES.ioc
>> include $(TOP)/configure/RULES
>>
>> to install the st.cmd to the binary install dir, at least. Additionally,
>> I set IOCS_APPL_TOP in CONFIG_SITE to the install location to let
>> envPaths look right. This works so far. Is this the way to go?
>>
>> In this context it might also be an idea to create a new template
>> directory for makeBaseApp.pl and point to it using the -T option. I'd
>> rather use the EPICS original ones, though, to have no adaption work
>> with future versions.
>>
>> (2) Another option I thought about would be to define a User Created
>> File type
>> (
>> https://epics.anl.gov/base/R3-15/6-docs/AppDevGuide/BuildFacility.html#x5-1570004.6.35
>> ), e.g. a FILE_TYPE IOCBOOT which would install this st.cmd to a
>> INSTALL_IOCBOOT = $(INSTALL_LOCATION)/iocBoot/.... directory, for
>> example. But same here: it's sort of 'proprietary' to do things like
>> these and such special solutions might be incompatible with future
>> versions of EPICS.
>>
>> Or do I think completely in the wrong direction and there is already a
>> mechanism  in place which to install the startup files to
>> INSTALL_LOCATION which I simply do not see?
>>
>> Any help or ideas as well as confirmation appreciated.
>>
>> Thanks,
>> Felix

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System