1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 <2021> 2022 2023 2024 | Index | 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 <2021> 2022 2023 2024 |
<== Date ==> | <== Thread ==> |
---|
Subject: | Re: EPICS latest stable version |
From: | Timo Korhonen via Tech-talk <tech-talk at aps.anl.gov> |
To: | Delahaye Olivier <olivier.delahaye at ganil.fr>, Ralph Lange <ralph.lange at gmx.de>, EPICS Tech Talk <tech-talk at aps.anl.gov> |
Cc: | Pillon Frank <frank.pillon at ganil.fr>, Zouaoui Djaid <djaid.zouaoui at ganil.fr>, Sebille Florian <florian.sebille at ganil.fr>, Patard Charles-Henri <charles-henri.patard at ganil.fr> |
Date: | Tue, 16 Feb 2021 15:23:03 +0000 |
Hi Olivier, ”the devil is in the details”… As Ralph pointed out, the process database is backwards compatible, to a very high degree. There may be a few changes here and there but I would be rather confident that this will not be a problem in the upgrade. However, the environment has had a lot of changes. With “environment” I mean systems that EPICS runs on, for instance VxWorks; if your installation has been running for a while, I guess you might still be at VxWorks versions
6 or even 5. Getting EPICS to compile on these platforms will have issues, and for version 5 it is practically impossible. So you would have to consider what to do with these, do you want to upgrade VxWorks, or maybe even need to
upgrade the hardware. Linux should be easier but you should still be prepared to do a number of the tedious small changes that Ralph mentioned. Channel Access should not be an issue, I would bet it works without issues in the same way it used to. And pvAccess can run parallel to Channel Access; the “tedious small changes” are not related to pvAccess, you can
even opt not to use it at all if you wish (this is certainly not a recommendation, though!) It is hard to estimate how long the upgrade would take. I assume that the Linux IOCs would be fairly easy to upgrade, but there is still work to get all the drivers and device support updated and tested, etc. I guess you would also like to update the client machines, even if old applications might work. But the big issue would be to decide what to do with your VxWorks IOCs – and that depends on what version you have been using so far, at least assuming you still want to continue using VxWorks. FWIW, we at ESS are not in regular beam production yet, but apart from a few early IOCs (Cryo, developed early) we are all on 7.0(.4) Best regards, Timo From:
Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Delahaye Olivier via Tech-talk <tech-talk at aps.anl.gov> Hi Ralph, Thx a lot for your quick answer. So If I understand well, we should better switch on 7.0 release. I would be quite confident about the stability since you mentionned half of the installations works now with 7.x version. Just a few more questions (sorry if questions are not clear, EPICS is quite new to me !). 1. What do you mean by tedious small changes ? We've got about 46 IOCs running on LINUX + 20 running on VxWorks. How long would you estimate for the upgrade ? 2. You mentionned pvAcess protocol has been reviewed on 7.0 version. Our equipments rely on Channel Acess protocol. Is it the same one ? Are the "tedious small changes" you are talking
about specifically related to the upgrade of the pvAccess protocol ? 3. You mentioned EPICS process database is backward-compatible. It's not clear to me the split between the different component in EPICS framework. But what is backward compatible and what
is not ? Because if everything is backward compatible, what do the "tedious small changes" consist of ? Thx a lot for your support, Olivier De : Ralph Lange <ralph.lange at gmx.de> Dear Olivier, I'm not getting into the "current" vs. "stable" discussion - that depends so much on your personal definition of these two terms that I can only lose. Let me answer it from a different angle.
·
3.14 is dead and doesn't even get bug fixes anymore. Time to move on. Not an option for new developments.
·
3.15 is supported and will get bugs fixed as they are discovered. For a new development inside an existing 3.15 context, this might be the appropriate option.
·
7.0 is being actively developed, provides the next generation network protocol pvAccess and keeps the robust 3.x IOC below. Recommended for any new developments. Such a migration will be an effort. No rocket science, for sure, but a lot of tedious small changes and testing, testing, testing. I would not do that more often than necessary. Most of the changes in EPICS 7 are in the new parts: the pvAccess protocol stack and its integration in the IOC. As long as you don't heavily rely on the pvAccess protocol, the migration risk is pretty low. The
EPICS Process Database code is robust and gets new features carefully in a backward-compatible manner, as always. Channel Access and its wide compatibility range will be the glue that keeps your system together. Migrating one system, even one IOC at a time should not be a big issue. Bottom line... Don't underestimate the task. You would need good reasons to not go to a recent 7.0 release. If I remember correctly, an audience poll at one of the last meetings showed that roughly half of the installations have EPICS 7 running somewhere. Can't be completely broken, then. Cheers, Préservons notre environnement, n’imprimez ce mail que si nécessaire. |