Re: building EPICSR3.14.8.1 on MacOSX 10.3/10.4

[Noboru Yamamoto <noboru.yamamoto@kek.jp>]

Eric Norum wrote:
> It appears that this mixup occurred when epicsStdlib.c was added.   What 
> about making the following change? 
>
> Index: epicsStdlib.c
> ===================================================================
> RCS file: 
> /net/phoebus/epicsmgr/cvsroot/epics/base/src/libCom/misc/epicsStdlib.c,v
> retrieving revision 1.1.2.1
> diff -u -r1.1.2.1 epicsStdlib.c
> --- epicsStdlib.c       12 Dec 2005 19:54:46 -0000      1.1.2.1
> +++ epicsStdlib.c       26 Dec 2005 18:22:40 -0000
> @@ -43,6 +43,9 @@
>      return 1;
> }
> +#ifdef epicsStrtod
> +# undef epicsStrtod
> +#endif
> epicsShareFunc double epicsStrtod(
>      const char *str, char **endp)
> {

> I'm quite surprised that this problem didn't show up during pre-release 
> testing.  I suspect that it will cause the same infinite recursion on 
> all POSIX targets. 

In the distribution, there are several place where I can find osdStrtod.h:

[noboru-g4:libCom/osi/os] noboru% find ./ -name osdStrtod.h -print
.//RTEMS/osdStrtod.h
.//WIN32/osdStrtod.h
.//cygwin32/osdStrtod.h
.//hpux/osdStrtod.h
.//posix/osdStrtod.h
.//vxWorks/osdStrtod.h

These files other than posix/osdStrtod.h include lines:

/*
 * epicsStrtod() for systems with broken strtod() routine
 */
epicsShareFunc double epicsShareAPI epicsStrtod(const char *str, char 
**endp);

However, the origiral distribution does not have this type of 
osdStrtod.h in the osi/os/Darwin directory and cases the infinite recursion.

Regards,

Noboru


> On Dec 26, 2005, at 12:14 AM, Noboru Yamamoto wrote:
>
> > Hi,
> >
> > I encountered a problem to run EPICS CA clients build with 
> > EPICSR3.14.8.1 under MacOSX 10.3/10.4.
> >
> >
> > In the osdStrtod.h under base-3.14.8.1/src/libCom/osi/posix, macro 
> > epicsStrtod is defined as:
> >
> > #define epicsStrtod strtod
> >
> > On the otherhand , the file epicsStdlib.c in base-3.14.8.1/src/libCom/ 
> > includes the definition of epicsStrtod. Because <epicsStrtod.h> is 
> > included before the definition of epicsStrtod function, this definition
> > acually defines strtod as a external funtion.
> >
> > The epicsStrtod calls a strtod function inside it. It ends up the
> > infinite calling sequence. For example, a simple "caget fred" command 
> > will never end in this environment.
> >
> > I'm not quite sure if this problem is related to my development 
> > environment( MacOSX 10.3/10.4 with gcc3, distributed by Apple . I 
> > don't use any fink command.)
> >
> > 1) Is there any other who encounter the same or similar problem with 
> > EPICS3.14.8.1?
> >
> > 2) there may be severay ways to avoid this phenomena, but which one do 
> > you like to take. For example, just remove the sentence
> >
> > #define epicsStrtod strtod
> >
> > or define a new macro like
> >
> > #define HAS_WORKING_STRTOD_IN_SYSTEM 0
> >
> > and skip the definition of epicsStrtod.
> >
> > Or better otherway?
> >
> > Thank you very much for your attension.
> >
> > Regards,
> >
> > Noboru Yamamoto
> > EPICS group/JPARC control group
> > KEK, JAPAN
>
> --
>
> Eric Norum <norume@aps.anl.gov <mailto:norume@aps.anl.gov>>
>
> Advanced Photon Source                     
>
> Argonne National Laboratory
>
> (630) 252-4793