sv_utf8_upgrade vs. bytes_to_utf8

[Bill Moseley]

I'm not clear from perlguts and perlapi how to make sure all my strings are utf8.

And I'm more confused about sv_utf8_upgrade vs bytes_to_utf8. 

Say I want to pass a list of strings passed into my xsub as an arrayref. 

Is there a problem with blindly calling sv_utf8_upgrade on every element of my AV?

For example, this code does not handle utf8:

void

foo( names )

    AV * names

    INIT:

        char ** name_list;

    CODE:

        Newx( name_list, av_len( names ) + 1, char * );

        for( int i=0; i <= av_len( names ); i++ )

            name_list[i] = SvPV_nolen( *av_fetch( names, i, 0) );

        RETVAL = foo( name_list, av_len( names) );

        SafeFree( name_list );

     OUTPUT:

         RETVAL

If my C function foo() expects all character data utf8-encoded is this the correct approach?

        for( int i=0; i <= av_len( names ); i++ ) {

            SV * name_sv = *av_fetch( names, i, 0);

            sv_utf8_upgrade( name_sv );
            name_list[i] = SvPV_nolen( name_sv );

         }

And is sv_utf8_upgrade a NOOP if the utf8 flag is already set?

Or is a better approach to use bytes_to_utf8()?   But, if I did that I would need to SafeFree() each string in my name_list[] array after calling my C function, correct?

--
Bill Moseley
mos...@hank.org