Re: record field max length?

[Jeong Han Lee via Tech-talk <tech-talk@aps.anl.gov>]

> I had 80 char buffers when INP/OUT could be 78 chars.
> ESS committed a fix to change the buffers to 256 in August 2018 to fit 
> EPICS 3.16 (?). This change is included in release 2.8.3.

In order to to fit "long pv names" generated by ESS site-specific 
program (PLCFactory). We are using 3.15.5 as the base version.
That is not "perfect", but based on Base 3.15.5, 256 is good enough to 
running IOC.
Best,
Han



> With the unlimited link size in EPICS 7 any fixed buffer size is 
> obviously not sufficient. More important parsing it without length check 
> is not save any more.
> I will change that.
> Dirk
>
>
> On 13.12.2018 16:55, Michael Davidsaver via Tech-talk wrote:
> > On 12/13/18 5:56 AM, Dirk Zimoch via Tech-talk wrote:
> > > Hi William,
> > >
> > > Even though the INP Length in EPICS 7 is not unlimited, StreamDevice 
> > > may nor be able to read it. 78 bytes was the effective limit of 
> > > INP/OUT in EPICS 3.14. Meanwhile I have changed my buffers from 80 to 
> > > 256 bytes but not yet to "unlimited" (which is a bigger change).
> > In case someone is motived to help.
> >
> > The buffers:
> >
> > https://github.com/paulscherrerinstitute/StreamDevice/blob/acf7efcff2e29a997a5c2d21629d7ef64f1e19de/src/StreamEpics.cc#L536-L542 
> >
> > Parsing with sscanf()
> >
> > https://github.com/paulscherrerinstitute/StreamDevice/blob/acf7efcff2e29a997a5c2d21629d7ef64f1e19de/src/StreamEpics.cc#L701-L753 
> >
> > > However your string is cur at 78 chars like in EPICS 3.14.
> > >
> > > I will check this with EPICS 7....
> > >
> > > Dirk
> > >
> > > On 13.12.2018 09:46, William Kirstaedter via Tech-talk wrote:
> > > > unlimited cant be the case, unfortunately. I'm running EPICS7 and 
> > > > facing the following error when starting the IOC:
> > > > 2018/12/13 09:42:51.409543 _main_ Missing ')' after substitutions 
> > > > 'readModule(EN,CRYVISIL:LEED:ENgain,CRYVISIL:LEED:ENoffset,CRYVISIL:LEED:ENvalue,LEED_01' 
> > > >
> > > > 2018/12/13 09:42:51.409621 _main_ CRYVISIL:LEED:getEN: Protocol 
> > > > parse error
> > > > 2018/12/13 09:42:51.409642 _main_ CRYVISIL:LEED:getEN: Record 
> > > > initialization failed
> > > > with these records and streamprotocol:
> > > >
> > > > # Energy
> > > > record(stringin, "$(P):$(R):getEN") {
> > > > field(DTYP,"stream")
> > > > field(INP,"@leed.proto 
> > > > readModule(EN,$(P):$(R):ENgain,$(P):$(R):ENoffset,$(P):$(R):ENvalue,$(P):$(R):ENmonitorU,$(P):$(R):ENmonitorI) 
> > > > $(PORT)")
> > > > }
> > > > record(ai, "$(P):$(R):ENgain") {
> > > > field(DESC,"Energy Gain")
> > > > field(EGU,"eV")
> > > > field(PREC,"2")
> > > > }
> > > > record(ai, "$(P):$(R):ENoffset") {
> > > > field(DESC,"Energy Offset")
> > > > field(EGU,"eV")
> > > > field(PREC,"2")
> > > > }
> > > > record(ai, "$(P):$(R):ENvalue") {
> > > > field(DESC,"Energy Value")
> > > > field(EGU,"eV")
> > > > field(PREC,"2")
> > > > }
> > > > record(ai, "$(P):$(R):ENmonitorU") {
> > > > field(DESC,"Energy Voltage Monitor")
> > > > field(EGU,"eV")
> > > > field(PREC,"2")
> > > > }
> > > > record(ai, "$(P):$(R):ENmonitorI") {
> > > > field(DESC,"Energy Current Monitor")
> > > > field(PREC,"2")
> > > > field(EGU,"A")
> > > > }
> > > > ------------
> > > > OutTerminator= CR;
> > > > InTerminator= "\r\000\n>";
> > > > readModule{
> > > > # $1 : Module Name -> REN (read energy)
> > > > out"R\$1";
> > > > # answer: R $1 r NUL n $1 (what out "typed in" plus return) 
> > > > $1|SP|GAIN|SP|OFFSET|SP|VALUE|SP|U-monitor|SP|I-monitor (the actual 
> > > > answer)
> > > > # >REN
> > > > # EN +0 +500 +500 +500.85 +1.5653E-06
> > > > in"R\$1\r\000\n\$1%(\$2)f%(\$3)f%(\$4)f%(\$5)f%(\$6)f";
> > > > }
> > > >
> > > > William Kirstaedter (PP&B)
> > > > Fritz-Haber-Institut der MPG
> > > > Faradayweg 4-6
> > > > 14195 Berlin
> > > > Tel: 030 8413 5405
> > > > Mail:kirstaedter@fhi-berlin.mpg.de
> > > >
> > > > Am 12.12.2018 um 17:29 schrieb Johnson, Andrew N.:
> > > > > On 12/12/18 8:59 AM, William Kirstaedter via Tech-talk wrote:
> > > > > > Short Question:
> > > > > >
> > > > > > Whats the maximum length of a record field? how many chars?
> > > > > >
> > > > > > record(ai,"record_name"){
> > > > > >
> > > > > >      field(DESC, "I_mean_this")
> > > > > >      field(INP, "how_long_can_this_be")
> > > > > >
> > > > > > }
> > > > > Longer answer:
> > > > >
> > > > > That depends on the field definition for the record type. DESC is 
> > > > > defined in dbCommon.dbd  (a kind of super-class used by all record 
> > > > > types) as:
> > > > > >          field(DESC,DBF_STRING) {
> > > > > >                  prompt("Descriptor")
> > > > > >                  promptgroup("10 - Common")
> > > > > >                  size(41)
> > > > > >          }
> > > > > That means it's a DBF_STRING and the size() entry says it can hold 
> > > > > up to 41 characters, but there must be a zero-byte terminator so 
> > > > > the limit is really 40 (which matches the Channel Access 
> > > > > dbr_string_t typedef).
> > > > > For something like the ZRST field of the mbbi record which defines 
> > > > > an enumeration string, the size is given as 26 which is the size of 
> > > > > the enumeration strings that Channel Access can send in a struct 
> > > > > dbr_gr_enum.
> > > > > An INP or OUT field is a link and for these the maximum length 
> > > > > depends on the version of EPICS you're running; it might be 40 for 
> > > > > older versions, 80, 256 for some or unlimited for EPICS-7.
> > > > > HTH,
> > > > >
> > > > > - Andrew
> > > > > --
> > > > > Arguing for surveillance because you have nothing to hide is no
> > > > > different than making the claim, "I don't care about freedom of
> > > > > speech because I have nothing to say." -- Edward Snowdon