Hello,

I thought you guys might find this interesting. It's a new 27 page document describing Unicode support in D2009.

http://dn.codegear.com/article/38980

------------------------------
Abstract: Learn more about the new Unicode support in Delphi 2009 and CodeGear RAD Studio 2009 in this white paper by Marco Cantù

Delphi and Unicode

One of the most relevant new features of Delphi 2009 is its complete support for the Unicode character set. While Delphi applications written exclusively for the English language and based on a 26-character alphabet were already working fine and will keep working fine in Delphi 2009, applications written for most other languages spoken around the world will have a distinct benefit by this change.

Learn more about Unicode in Delphi 2009 and CodeGear RAD Studio 2009 in this white paper.
------------------------------

Regards,
- Graeme -

_______________________________________________
fpGUI - a cross-platform Free Pascal GUI toolkit
http://opensoft.homeip.net/fpgui/

Seeing that I don't own D2009 and only read about it's Unicode support I found some of the information interesting - and it was things we argued about in this mailing list.

For example:

1...
Length() returns the bytes for UTF8String
but Length() returns the elements (what we know as characters) for String or UTF16 strings.
Length() also returns bytes for AnsiString.

--------------------

        var
          str8: Utf8String;
          str16: string;
        begin
          str8 := 'Cantù';
          Memo1.Lines.Add ('UTF-8');
          Memo1.Lines.Add('Length: ' + IntToStr (Length (str8)));
          Memo1.Lines.Add('5: ' + IntToStr (Ord (str8[5])));
          Memo1.Lines.Add('6: ' + IntToStr (Ord (str8[6])));
          str16 := str8;
          Memo1.Lines.Add ('UTF-16');
          Memo1.Lines.Add('Length: ' + IntToStr (Length (str16)));
          Memo1.Lines.Add('5: ' + IntToStr (Ord (str16[5])));

As you might expect, the str8 string has a length of 6 (meaning 6 bytes), while the str16
string has a length of 5 (meaning 10 bytes, though). Notice that Length invariably returns the
number of string elements, which in case of variable-length representations don't match the
number of Unicode code points represented by the string. This is the output of the program:
UTF-8
Length: 6
5: 195
6: 185
UTF-16
Length: 5
5: 249

--------------------

2... TStrings can now take an encoding parameter to specify how it should load or save files.

-----------------------------
STREAMING TSTRINGS
The ReadFromFile and WriteToFile methods of the TStrings class can be called with
an encoding. If you write a string list to text file without providing a specific encoding, the class
will use TEncoding.Default, which uses the internal DefaultEncoding in turn extracted at the first occurrence by the current Windows code page. In other words, if you save
a file you'll get the same ANSI file as before.
Of course, you can also easily force the file to a different format, for example the UTF-16 format:

            Memo1.Lines.SaveToFile('test.txt',  TEncoding.Unicode);
-----------------------------

anyway, there are a lot more interesting facts in this document. Well worth reading to get a better understanding of unicode.

Regards,
- Graeme -

_______________________________________________
fpGUI - a cross-platform Free Pascal GUI toolkit
http://opensoft.homeip.net/fpgui/

In our previous episode, Graeme Geldenhuys said:

This is all information that is already on the blogs since July. Note that Tcharacter is a sealed class, something that FPC doesn't support yet.

The whole tencoding/tcharacter is a bastard-class stuff seems to be out of .NET compatibility (as noted in the document), but Borland changed course of its .NET efforts after Tiburon. Sigh.

No Length for String will return the number of Code Units (the number of WideChar in UnicodeString case). When there's surrogate pairs it will differ the number of Code Points (Characters) and Code Units. See the excerpt:

"
A way to create a string with surrogate pairs is to use the ConvertFromUtf32 function that
returns a string with the surrogate pair (two WideChar) in the proper circumstances, like the
following:

var
str1: string;
begin
str1 := 'Surr. ' + ConvertFromUtf32($1D11E);

Now if you ask for the string length, you'll get 8, which is the number of WideChar, but not the
number of logical Unicode code points in the string. If you print the string you get the proper
effect (well, at least Windows will generally show one square block as placeholder of the
surrogate pair, rather than two).
"

Well, with exclusion of the "class helper for TStrings" (notable is that they call it a hack themselves :) the design looks rather clean. Since each string stores its element size, both ansi and unicode strings are probably handled with common set of procedures, avoiding RTL size bloat.

And they explain why there is no compiler option for switching back and forth.

Unfortunately, the article does not provide information about how things like Pos() and Copy() work with utf8 strings. However, one may understand words "utf-8 support is more limited than utf-16" as they continue to work with elements (bytes).

Regards,
Sergei

I also like the design since is flexible enough to allow the programmer work with different encodings.

Here ( http://www.jacobthurman.com/?p=30 see comments) there's an explanation about those functions. Basically they will handle Code Units and not Code Points (characters)

However, one may understand words "utf-8 support is more limited than utf-16" as they continue to work with elements (bytes).

Yes. This is a good decision also IMO.

Luiz

2008/11/21 Graeme Geldenhuys :

Memo1.Lines.SaveToFile('test.txt', TEncoding.Unicode);

I am not agree to make TStrings responsible of converting to another encoding, it is enough for us save and load the text as it in the same application (TStringList).
We need to convert when need to read result text from another applications to send it, there you need some functions to make converting pass the TStrings as params.