Minutes from the 11/17/04 LCLS SLC IOC meetings.

Next Meeting is 11/24/04, 9:30am, second floor.

(1) Progress (Message and Database):
    Diane is done with the slcExec and startup/shutdown design documents and
    is working on the message service document.  Debbie is working on her
    database document.  They'll copy the .doc and .html files to the SLC IOC 
    web area when finished. Note that their review presentations are now
    slightly outdated due to review comments and other post-review thoughts.
    This is OK - no need to keep the presentations up-to-date. Kristi will 
    forward an example of a controls design document to us.  Both Debbie and 
    Diane need until at least the end of the year to finish implementation.

(2) Progress (iocCmlog):
    James is working on it - the design has changed due to review comments.  
    He will have a review on Dec 2 at 1:30 in the MCC conference room.

(3) Progress (device control):
    Kristi is looking at the reqts done so far by Dayle and will add to them.
    It's possible she'll be ready for a func reqts review on Dec 15 or 16.

(4) Progress (async utilities and CSTR update):
    Steph reports no progress.

(5) New web location:
    Kristi has set up a new 1 GB area for our web documents:
    /afs/slac/www/grp/lcls
    We'll move our pages and documents to a subdirectory there when she
    is ready.  She has developed a simple template that all pages will use
    to keep a similar look-and-feel.

(6) Design Review Decisions:
    Several design review comments required decisions.  We've made these
    decisions:
    (a) Tasks will be suspended on fatal errors.  The exception is where
        an EPICS utility called by a task forces an exit on a fatal error.  
    (b) Diane will decrease the number of copies.  Note that we will not
        be supporting 64 bit processors so the message header can be 
	converted in-situ.
    (c) We will plan for dbstop and dbrestart functionality.  Tasks setting up 
        or using the SLC database will not exit but will reinitialize.  
	The SLC exec will need to set a flag in the threads structure for all 
	tasks to honor.
    (d) The database version will be added to the dblist list structure and
        the dbl* routines will use it to make sure the database hasn't
	changed out from under them.
    (e) Debbie will create her own hashing utilities, based on gphash,
        for the database dictionary.
    (f) Supertype 1 dbedits - we will allow them only if a new global flag
        is set.  This flag will default to not-allowed.
    (g) We will add comments to the how-to-add-a-micro document on how to
        assure that a unique micro name is used when a new IOC is added.
	The new SLC-aware IOC must be updated (carefully) in
        REF_DBSFILE:MICRONAME.DAT in CMS. Tony says that
        REF_IOBSIC:MSG_CRMICRONAME_READ_FILE.FOR (which reads
        MICRONAME.DAT) rejects duplicate micro name (previously he
        checked only against duplicate bitid). 

(7) Design Review Comments - Later:
    The following tasks will be done later:
    (1) Add another proxy if needed.  Significant work.  At some point,
        we'll be able to get an idea of how a single proxy will handle
	so many new micros and we can decide if a second proxy is needed.
    (2) Increase the maximum number of micros in the micro-active-mask.
        Significant Alpha work. At some point, we'll have a better idea of 
	how many new micros are needed for LCLS.
    (3) Security for channel access request for SLC restart.
    (4) Common startup file for all SLC IOCs.
    (5) Comments in SLC IOC and SLC micro source to reference each other.
    (6) Adding "primary secondary" to data dictionary for the ALL* unit
        case.  We'll see if the other tasks will do many of these besides
	during initialization.

(8) Design Review Comments - Never:
    The following tasks are not required by the LCLS project:
    (1) SCP to use the CMLOG browser.
    (2) SLC IOC to get primary.map database definition from Oracle 
        instead of file.
    (3) Add extra logic on the Alpha (per application) to check that
        a micro database update is finished.
    (4) 5 character unit names.
 
(9) File Writing by IOC Shell
    Kristi wonders how writing large NFS files by IOC shell affects the 
    performance of other tasks on the SBC.  Should we remove file writing
    from IOC shell commands?