Re: Out_File , excess line

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On 2016-01-27 10:26, Simon Wright wrote:
> comicfanzine@gmail.com writes:
>
>> But , when we try to open the file manually , the text editor print :
>>
>> " This is not a text file . Type unknown ."
>>     PACKAGE ST_IO RENAMES Ada.Streams.Stream_IO ;
>>
>>     test_file : ST_IO.File_type ;
>>
>>     Input_Stream : ST_IO.Stream_Access ;
> [...]
>>     ST_IO.Create( File => test_file ,
>>                 Mode => ST_IO.Out_File ,
>>                 Name => "test" );
>>
>>      Input_Stream := ST_IO.Stream( test_file );
>>
>>      for j in 1..case_array'Length loop
>>        Unbounded_String'Write( Input_Stream , case_array(j) );
>>     end loop;
>
> 'Input_Stream' is a very odd name for a stream you're going to *output*
> to.
>
> What's happening is that (reasonably, though not as far as I can see
> discussed by the standard) GNAT treats Unbounded_String'Write as if an
> unbounded string is a discriminated type whose discriminants have
> defaults; that is, the discriminants (the array bounds) are output to
> the stream first, followed by the characters of the string. (This is
> what happens for String'Output; String'Write doesn't include the bounds;
> is the Unbounded_String'Write behaviour a bug or a feature?).
>
> So, if you Unbounded_String'Write "foo", you will get
> - 4 bytes containing binary 1, the first index of the content
> - 4 bytes containing binary 3, the last index of the content
> - 3 bytes of content, 'f' 'o' 'o'.
>
> Which looks nothing like a text file.
>
> You could try
>
>    String'Write (Input_Stream, To_String (case_array (j)) & ASCII.LF);

This one of many reasons why no Unbounded_String should ever be used for 
any text processing. That includes text buffers of course. Using 
standard containers of Unbounded_String is probably least efficient 
possible way to implement a text buffer.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de