Re: EPICS Base Release-Candidate Double Feature!

[Matt Rippa <mrippa@gemini.edu>]

Hi Andrew,

On Thu, Nov 17, 2016 at 1:42 PM, Andrew Johnson <anj@aps.anl.gov> wrote:

Hi Matt,

The RSET::get_value() routine has not been used for many many years,
even the 3.13 and probably 3.12 IOC core code never called it or used
the struct valueDes type. I removed the struct valueDes definition
because it wasn't being used anywhere in the Base code, and to sniff out
any remaining record types that might be implementing a get_value
routine, which can be deleted. You'll need to use a NULL for the RSET
function pointer for get_value instead.

Your other error is because the field name NOT is no longer lower-cased
when converting it to the record member name in 3.15. The word 'not' is
a C++ keyword so can't be used as a struct member name. Change your code
to use ->NOT instead of ->not.

Thanks for that. I've updated half a dozen of our custom records which are all very old.

I can't seem to track down a change to dbGetLinkValue() which is now missing from R3.15.

In R3.14.12.4 it's defined here:

dbAccessDefs.h: epicsShareFunc long epicsShareAPI dbGetLinkValue(

    struct link *,short dbrType,void *pbuffer,long *options,long *nRequest);

I can't find any related release notes.

There may be a number of places in your applications where you can
replace the genSub with an aSub record type instead, which has various
bugs fixed. Search the release notes for 3.14 to find a description of
that record type.

Would be great to do this. The "number of places" is daunting. 

-Matt

 

HTH,

- Andrew

On 11/17/2016 05:25 PM, Matt Rippa wrote:
> I'm testing Gemini's genSubRecord.c against R3.15.5-rc1. Looks like our
> RSET implementation is now obsolete as of R3.15.3 (See:
> http://www.aps.anl.gov/epics/base/R3-15/3-docs/RELEASE_NOTES.html
> RSET::get_value()).
>
> Is there a porting guide to Record support for R3.15 ?
>
> Thank you,
> Matt
> ---------------------
> Build Error:
>
> make[3]: Entering directory
> `/home/mrippa/work/R3.15.5/support/geminiRec/geminiRecApp/src/O.linux-x86_64'
> /usr/bin/gcc  -D_GNU_SOURCE -D_DEFAULT_SOURCE            -D_X86_64_
>  -DUNIX  -Dlinux     -O3 -g   -Wall      -mtune=generic     -m64 -fPIC
> -I. -I../O.Common -I. -I. -I.. -I../../../include/compiler/gcc
> -I../../../include/os/Linux -I../../../include
> -I/gem_sw/epics/R3.15.5-rc1/base/include/compiler/gcc
> -I/gem_sw/epics/R3.15.5-rc1/base/include/os/Linux
> -I/gem_sw/epics/R3.15.5-rc1/base/include        -c ../genSubRecord.c
> ../genSubRecord.c:607: warning: ‘struct valueDes’ declared inside
> parameter list
> ../genSubRecord.c:607: warning: its scope is only this definition or
> declaration, which is probably not what you want
> ../genSubRecord.c: In function ‘get_value’:
> ../genSubRecord.c:612: error: dereferencing pointer to incomplete type
> ../genSubRecord.c:613: error: dereferencing pointer to incomplete type
> ../genSubRecord.c:614: error: dereferencing pointer to incomplete type
> ../genSubRecord.c: In function ‘findField’:
> ../genSubRecord.c:1076: error: ‘genSubRecord’ has no member named ‘not’
> ../genSubRecord.c:1080: error: ‘genSubRecord’ has no member named ‘not’
> ../genSubRecord.c:1082: error: ‘genSubRecord’ has no member named ‘not’
> make[3]: *** [genSubRecord.o] Error 1
> make[3]: Leaving directory
> `/home/mrippa/work/R3.15.5/support/geminiRec/geminiRecApp/src/O.linux-x86_64'
>
> Relevant legacy code:
> genSubRecord.c
> #include<stdlib.h>
> #include<stdio.h>
> #include<string.h>
>
> #include<alarm.h>
> #include<dbDefs.h>
> #include<dbEvent.h>
> #include<dbAccess.h>
> #include<dbFldTypes.h>
> #include<errMdef.h>
> #include<recSup.h>
> #include<devSup.h>
> #include <special.h>
> #include <registryFunction.h>
> #include <epicsExport.h>
> #include <recGbl.h>
>
> ...
> rset genSubRSET={
> RSETNUMBER,
> report,
> initialize,
> init_record,
> process,
> special,
> get_value,
> cvt_dbaddr,
> get_array_info,
> put_array_info,
> get_units,
> get_precision,
> get_enum_str,
> get_enum_strs,
> put_enum_str,
> get_graphic_double,
> get_control_double,
> get_alarm_double };
> epicsExportAddress(rset,genSubRSET);
> ...
>
> static long get_value( genSubRecord *pgsub, struct valueDes *pvdes )
> {
> #if DEBUG
>     printf("Calling get_value...\n");
> #endif
>     pvdes->no_elements = 1;
>     pvdes->pvalue      = (void *)(&pgsub->val);
>     pvdes->field_type  = DBF_LONG;
>     return(0);
> }
>
>
>
> On Sun, Nov 13, 2016 at 8:38 AM, <freddie.akeroyd@stfc.ac.uk

> <mailto:freddie.akeroyd@stfc.ac.uk>> wrote:
>
>     Hi Andrew,
>
>     Sorry, I didn't refer to this in my original reply, but I had also
>     changed the first line from "#if defined(WIN32)"  to  "#if
>     defined(_WIN32)" as "WIN32" is not defined by Visual Studio (it is
>     by MinGW though in addition to  _WIN32). So the 3.15 test program it
>     was taken from builds with VS 2015 as it never actually tries to
>     call  _set_output_format() .
>
>     Regards,
>
>     Freddie
>
>     -----Original Message-----
>     From: Johnson, Andrew N. [mailto:anj@aps.anl.gov
>     <mailto:anj@aps.anl.gov>]
>     Sent: 12 November 2016 17:03
>     To: Akeroyd, Freddie (STFC,RAL,ISIS)
>     Cc: tech-talk@aps.anl.gov <mailto:tech-talk@aps.anl.gov>
>     Subject: Re: EPICS Base Release-Candidate Double Feature!
>
>     Hi Freddie,
>
>     Thanks, I will commit that change on Monday. I copied that code
>     including the conditional from one of the test programs in 3.15
>     src/ioc/db/test though, so I'm wondering how you didn't see a build
>     failure with VS2015 there - could you investigate please?
>
>     Regards,
>
>     - Andrew
>
>     --
>     Sent from my iPad
>
>     > On Nov 12, 2016, at 8:06 AM, "freddie.akeroyd@stfc.ac.uk
>     <mailto:freddie.akeroyd@stfc.ac.uk>" <freddie.akeroyd@stfc.ac.uk
>     <mailto:freddie.akeroyd@stfc.ac.uk>> wrote:
>     >
>     > Hi Andrew,
>     >
>     > Your patch does indeed fix the formatting  problem present in
>     earlier versions of Visual Studio, but it appears that when
>     Microsoft corrected this issue in Visual Studio 2015 they decided to
>     remove the  _set_output_format()  function at the same time! So I
>     just needed to adjust the test to:
>     >
>     > #if defined(_WIN32)
>     > #if    (defined(_MINGW) &&  __MSVCRT_VERSION__ >= 0x0800) ||  \
>     >              (defined(_MSC_VER) && _MSC_VER < 1900)
>     >    _set_output_format(_TWO_DIGIT_EXPONENT);
>     > #endif
>     > #endif /* defined(_WIN32) */
>     >
>     > and then it works OK with both VS 2010 and 2015 for     testPlan(163)
>     >
>     > Regards,
>     >
>     > Freddie
>     >
>     > -----Original Message-----
>     > From: Andrew Johnson [mailto:anj@aps.anl.gov <mailto:anj@aps.anl.gov>]
>     > Sent: 11 November 2016 20:43
>     > To: Akeroyd, Freddie (STFC,RAL,ISIS); tech-talk@aps.anl.gov
>     <mailto:tech-talk@aps.anl.gov>
>     > Subject: Re: EPICS Base Release-Candidate Double Feature!
>     >
>     > Hi Freddie,
>     >
>     >> On 11/10/2016 05:52 PM, freddie.akeroyd@stfc.ac.uk

>     <mailto:freddie.akeroyd@stfc.ac.uk> wrote:
>     >> epicsStdioTest.c plans 166 tests on Windows but 163 otherwise, if I
>     >> build using Visual Studio 2010 instead 166 tests passed OK. On
>     >> further investigation it seems the issue occurs within
>     >> testEpicsSnprintf() and is due to Microsoft changing (fixing) the
>     >> formatting  of "%8.2e" by sprint():
>     >>
>     >> In Visual Studio 2010 this generates    1.23e+004
>     >> In Visual Studio 2015 this generates    1.23e+04
>     >
>     > Thanks for the report. Does this patch make the test pass on both
>     compilers?
>     >
>     > === modified file 'src/libCom/test/epicsStdioTest.c'
>     > --- src/libCom/test/epicsStdioTest.c    2012-02-03 22:12:06 +0000
>     > +++ src/libCom/test/epicsStdioTest.c    2016-11-11 20:40:36 +0000
>     > @@ -39,6 +39,11 @@
>     >     int size;
>     >     int rtn, rlen;
>     >
>     > +#if defined(WIN32) && \
>     > +    (!defined(_MINGW) || __MSVCRT_VERSION__ >= 0x0800)
>     > +    _set_output_format(_TWO_DIGIT_EXPONENT);
>     > +#endif
>     > +
>     >     sprintf(exbuffer, format, ivalue, fvalue, svalue);
>     >     rlen = strlen(expected)+1;
>     >
>     > @@ -122,11 +127,7 @@
>     >
>     > MAIN(epicsStdioTest)
>     > {
>     > -#ifdef _WIN32
>     > -    testPlan(166);
>     > -#else
>     >     testPlan(163);
>     > -#endif
>     >     testEpicsSnprintf();
>     >     testStdoutRedir("report");
>     >     return testDone();
>     >
>     >
>     > - 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
>
>

--
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