https://community.intel.com/t5/Intel-Fortran-Compiler/Pointing-etiquette/m-p/772970#M23503 <DIV>You cannot have T(:)%component(:) notation <STRONG>anywhere</STRONG>according tothe Standard. I can't quoteexact reasons, butI think that 1) itcan lead to ambiguities, 2) in general case, an array/pointer descriptor cannot be always constructed to match the resulting memory layout.</DIV> <DIV></DIV> <DIV>Thus, you have to resort to DO-loops.</DIV> <DIV></DIV> <DIV>Jugoslav</DIV> Fri, 09 Apr 2004 16:21:36 GMT Jugoslav_Dujic 2004-04-09T16:21:36Z https://community.intel.com/t5/Intel-Fortran-Compiler/Pointing-etiquette/m-p/772969#M23502 <DIV>Under CVF6.6B</DIV> <DIV>This works:</DIV> <DIV></DIV> <DIV><STRONG>type symbolType<BR />real, pointer:: a(:),b(:)<BR />end type symbolType<BR />type(symbolType):: symbol</STRONG></DIV> <DIV><BR /><STRONG> symbol.a =&gt; symData(:,1)<BR /> symbol.b =&gt; symData(:,2)</STRONG></DIV> <DIV><STRONG></STRONG></DIV> <DIV></DIV> <DIV>But its extension to an array does not:</DIV> <DIV><STRONG>type symbolType<BR />real, pointer:: a(:),b(:)</STRONG></DIV> <DIV><STRONG>end type symbolType<BR />type(symbolType):: symbol(4)</STRONG></DIV> <DIV><BR /><STRONG> symbol(:).a(:) =&gt; symSetData(:,1,:) !or (:,:,1)<BR /> symbol(:).b(:) =&gt; symSetData(:,2,:)<BR /> </STRONG></DIV> <DIV>Where the intent is the first (:) on the lhs should point to the first onthe rhs etc, Is thereANY way that this can be achieved?</DIV> <DIV></DIV> <DIV>I have tried</DIV> <DIV><STRONG>type(symbolType),pointer:: symbol(:)</STRONG></DIV> <DIV><STRONG>type(symbolType),allocatable:: symbol(:) !followed by allocation</STRONG></DIV> <DIV>and a number of other variations, But none compile.</DIV> <DIV></DIV> <DIV></DIV> <DIV>Thanks, </DIV> <DIV>Tim</DIV> Fri, 09 Apr 2004 01:45:07 GMT https://community.intel.com/t5/Intel-Fortran-Compiler/Pointing-etiquette/m-p/772969#M23502 rahzan 2004-04-09T01:45:07Z https://community.intel.com/t5/Intel-Fortran-Compiler/Pointing-etiquette/m-p/772970#M23503 <DIV>You cannot have T(:)%component(:) notation <STRONG>anywhere</STRONG>according tothe Standard. I can't quoteexact reasons, butI think that 1) itcan lead to ambiguities, 2) in general case, an array/pointer descriptor cannot be always constructed to match the resulting memory layout.</DIV> <DIV></DIV> <DIV>Thus, you have to resort to DO-loops.</DIV> <DIV></DIV> <DIV>Jugoslav</DIV> Fri, 09 Apr 2004 16:21:36 GMT https://community.intel.com/t5/Intel-Fortran-Compiler/Pointing-etiquette/m-p/772970#M23503 Jugoslav_Dujic 2004-04-09T16:21:36Z