Re: building labCA on base 3.15.2, missing tsDefs.h

[Till Straumann <strauman@slac.stanford.edu>]

On 09/08/2015 02:43 PM, Gierman, Stephen wrote:
> Hi Till,
>
> Thanks for labCA-3.5.0.  I've built and tested it on windows-x64 and it mostly seems to be working.  I get an odd message when I build:
>
> C:\epics\R3.15.2\extensions\src\labca_3_5_0>make > make_labca_3_5_0.txt
> process_begin: CreateProcess(NULL, sed -n -e "s/\(.*[^b]subsection.*[{][ \t]*\)\(lca[a-zA-Z0-9]*\).*/\2/gp" manual.tex, ...) failed.
Thanks. That's luckily benign (safe to ignore) but nevertheless a bug in 
the makefile. The above command
should not have to be executed unless documentation is rebuilt from the 
.tex file. I usually create docs
during the release process so there is no need for 'sed' -- which may 
not be available under windows.

> And I may have a problem with the new lcaGetEnumStrings().  Here's what happens in matlab R2015a:
>
> > > lcaGetEnumStrings('LTD-Q1D:Control')
> Invalid MEX-file 'C:\epics\R3.15.2\extensions\src\labca_3_5_0\bin\windows-x64\labca\lcaGetEnumStrings.mexw64': The specified
> procedure could not be found.
Hmm - I'm not able to reproduce this. Works fine for me -- but the only 
matlab I have access to is
2012a/linux-x86 and 2014b/linux-x86_64.

Do the other entry points work (e.g., 'lcaGet()') ?

- Till
> Steve
>
>
> -----Original Message-----
> From: Till Straumann [mailto:strauman@slac.stanford.edu]
> Sent: Tuesday, September 08, 2015 12:56 PM
> To: Gierman, Stephen; 'tech-talk@aps.anl.gov'
> Subject: Re: building labCA on base 3.15.2, missing tsDefs.h
>
> Yeah. tsDefs is cruft from the old days.
>
> I just put up a new release (labca-3.5.0) which does away with tsDefs and TS_STAMP (replaced by epicsTime.h and epicsTimeStamp, respectively).
>
> This release also features a new call 'lcaGetEnumStrings()' which allows you to retrieve the set of possible values for enum-type PVs.
>
> Available as usual from
>
> http://www.slac.stanford.edu/~strauman/labca/
>
> - Till
>
> On 09/07/2015 03:23 PM, Gierman, Stephen wrote:
> > Hi,
> >
> > I'm building labCA (version 3.4.2) on base-3.15.2 for 64-bit windows.  The build fails in the 'glue' directory when it cannot find the include file 'tsDefs.h'.  Indeed, I do not find tsDefs.h in base-3.15.2, but I do find it in base-3.14.12.2, though it seems to be deprecated there.  Here are the last few lines of the build:
> >
> > cl -EHsc -GR            -DMATLAB_APP     -nologo -D__STDC__=0 -D_CRT_SECURE_NO_DEPRECATE -D_CRT_NONSTDC_NO_DEPRECATE   -Ox -GL -Oy-   -W3 -w44355        -MD -DEPICS_BUILD_DLL -DEPICS_CALL_DLL -TP  -I. -I../O.Common -I. -I. -I.. -I../../include/compiler/msvc -I../../include/os/WIN32 -I../../include -I../../../../../base/include/compiler/msvc -I../../../../../base/include/os/WIN32 -I../../../../../base/include           -I../../ezca -IC:\matlab\R2015a/extern/include    -c ../ini.cc
> > ini.cc
> > ..\multiEzca.h(32) : fatal error C1083: Cannot open include file:
> > 'tsDefs.h': No such file or directory
> > make[2]: Leaving directory `C:/epics/R3.15.2/extensions/src/labca_3_4_2/glue/O.windows-x64'
> > make[1]: Leaving directory `C:/epics/R3.15.2/extensions/src/labca_3_4_2/glue'
> >
> > I'm not sure what to do next ...
> >
> > Steve Gierman