Re: [AMBER] Hack to install AmberTools20 on macOS

From: Gustaf Olsson <gustaf.olsson.lnu.se>
Date: Tue, 20 Apr 2021 06:44:36 +0000

I suppose this makes me a stickler here but you’ve actually helped solve a very likely related issue on February 9th

    Sorry that I was looking at the development branch code. Still, try adding
    the following two lines right after line 199 above:

    int yyerror( char *sStr );
    int yylex();

    ...thx…dac

The workaround suggested at that time was to not edit the files after build failure but rather edit the parser.y file:

    gsed -i '/\extern int yyparse();/a int yyerror( char *sStr );\nint yylex();' AmberTools/src/leap/src/leap/parser.y

After which compilation was successful. I suppose this point got lost in the discussion regarding using the homebrew gfortran compiler.

Best regards
// Gustaf


> On 19 Apr 2021, at 15:27, David A Case <david.case.rutgers.edu> wrote:
>
> On Sun, Apr 18, 2021, Otto Sievert wrote:
>>
>> Today I helped someone install AmberTools20 on his macOS computer. I wanted
>> to report a problem with this, and my hacky work-around. At a minimum, this
>> work-around might help others.
>>
>> The issue: after installing prerequisites (Xcode command line tools,
>> gfortran 8.2, cmake) the build fails to compile the file
>> build/AmberTools/src/leap/src/leap/y.tab.c file (a generated source file)
>> because the C compiler requires a forward declaration of the two functions
>> yylex() and yyerror() before they are referenced.
>>
>> My hacky work-around: after getting that build failure, I edited the
>> generated y.tab.c file, added these two forward declaration lines near the
>> top of the file, and re-ran "make":
>> int yylex();
>> int yyerror(char *sStr);
>
> Thanks for the report. This is fixed in the development version, and will
> be in AmberTools21, which will be released soon.
>
> It's odd that no one else has reported this (which is why we haven't
> released an update patch). It used to be flagged as a warning for some
> compilers, but it seems that other compilers are now issuing an error rather
> than a warning, and are no longer allowing implicit declarations.
>
> ....regards....dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 20 2021 - 00:00:02 PDT
Custom Search