Re: PREC

[Marty Kraimer <mrk@aps.anl.gov>]

IOC database access converts DBF_FLOAT to DBR_STRING by calling 
cvtFloatToString.
Similarly DBF_DOUBLE is converted to DBR_STRING by calling 
cvtDoubleToString.
Both accept precision as an argument.

In 3.14 they are in <base>/src/libCom/cvtFast/cvtFast.c

For moderate size values precision means number of decimal places.  See 
the code for how it handles large numbers or precision > 8.

Early versions of vxWorks had a very slow implementation of %g format 
conversion. It could only do about 100 conversions per second on a 
68020. Thus cvtFast.c was created.

Note that cvtFast has a number of usefull converrsion routines:


       cvtFloatToString(float value, char *pstring, unsigned short 
precision);
       cvtDoubleToString(double value, char *pstring, unsigned short 
precision);
       cvtFloatToExpString(float value, char *pstring, unsigned short 
precision);
       cvtDoubleToExpString(double value, char *pstring, unsigned short 
precision);
       cvtFloatToCompactString(float value, char *pstring, unsigned 
short precision);
       cvtDoubleToCompactString(double value, char *pstring, unsigned 
short precision);
       cvtCharToString(signed char value, char *pstring);
       cvtUcharToString(unsigned char value, char *pstring);
       cvtShortToString(short value, char *pstring);
       cvtUshortToString(unsigned short value, char *pstring);
       cvtLongToString(epicsInt32 value, char *pstring);
       cvtUlongToString(epicsUInt32 value, char *pstring);
       cvtLongToHexString(epicsInt32 value, char *pstring);
       cvtLongToOctalString(epicsInt32 value, char *pstring);


Dirk Zimoch wrote:

> Hi all,
>
> we all know the PREC field of many records. Unforunately, its purpose 
> is not documented very precisely. The Application Developer's Guide 
> says in chapter 3.5.1 (EPICS Overview - Channel Access - Client 
> Services):
>
> "In addition to requesting process variable values, any combination of 
> the following additional information may be requested:
> [...]
>   * Precision: Precision with which to display floating point numbers."
>
> In the Record Reference Manual, one can read e.g. in the ai chapter:
>
> "The PREC field determines the floating point precision with which to 
> display VAL."
>
> But what does this exactly mean? I think, the most intuitive sense is 
> to format the value like "%.<PREC>f". At least this is the default 
> behaviour of medm when using a text monitor. But why not like 
> "%.<PREC>g" or "%.<PREC>e"? And why does neither caget nor camonitor 
> honor PREC? And what about negative PREC values?
>
> I suggest the following:
> PREC >= 0 means to format the value like printf("%.*f",PREC,VAL);
> PREC < 0  means to format the value like printf("%.*e",-PREC,VAL);
>
> All tools including caget and camonitor should honor PREC.
>
> Any comments?
>
> Dirk