Some basic questions

[slot]

Is there any problem to use "std::string" with unicode strings?

When using std::string, does it have to be initialized? Is the following
code OK?

string s;
S = "This is a test string";


Is it OK to have a container (for example, list), as global object whose
size and elements increases at run time?

Thanks!

 

[Victor Bazarov]

Is there any problem to use "std::string" with unicode strings?

There may be. Unicode character sequences can contain many 0 chars,
and that can present a problem in using the 'c_str()' member function
if you expect it to return a zero-terminated c-string.

When using std::string, does it have to be initialized?

If you don't provide an initialiser, the string will be default-
initialised, which means that it will be empty.

Is the following
code OK?

string s;
S = "This is a test string";

No, it's not OK. C++ is case-sensitive. If you declare a variable
called 's', assigning to 'S' later doesn't make sense (unless you also
declared a variable with the name 'S').

Is it OK to have a container (for example, list), as global object whose
size and elements increases at run time?

Yes. Although, in many cases global objects can and should be avoided.

Victor

 

[Alex Vinokur]

There may be. Unicode character sequences can contain many 0 chars,
and that can present a problem in using the 'c_str()' member function
if you expect it to return a zero-terminated c-string.

Here is a link to relevant discussion:
http://groups.google.com/[email protected]

[snip]

 

[Pete Becker]

There may be. Unicode character sequences can contain many 0 chars,
and that can present a problem in using the 'c_str()' member function
if you expect it to return a zero-terminated c-string.

Well, yes and no. (Sorry, Victor, I seem to be in the mood for quibbling
with you today <g>). A std::basic_string<wchar_t> works fine for Unicode
if wchar_t is large enough to hold individual Unicode characters. In
that case you only get zeroes if you put them there. If a wchar_t isn't
large enough you typically use one of the UTF-x encodings. The easiest
one to use portably is UTF-8, becuase you can always hold 8-bit
characters in a std::basic_string<char> (also known as std::string). A
UTF-8 encoded Unicode string only has zero values when you put them
there; the encoding doesn't generate spurious zeroes.

On the other hand, if you try to force 16-bit Unicode values into a
std::string by simply splitting each Unicode value into two 8-bit
pieces, you'll indeed get lots of zeroes. UTF-8 is designed to avoid
that, at the cost of sometimes requiring more than two bytes to encode a
16-bit value.

 

[Victor Bazarov]

There may be. Unicode character sequences can contain many 0 chars,
and that can present a problem in using the 'c_str()' member function
if you expect it to return a zero-terminated c-string.

Well, yes and no. (Sorry, Victor, [...]

Don't worry 'bout it. I am in no sense an expert on Unicode and
sincerely hoped that somebody would correct me and expand on my so
amateurish attempt at an explanation...

V

 

[Ioannis Vranos]

Is there any problem to use "std::string" with unicode strings?

Yes std::string is for chars. Try std::wstring instead (which is for
wchar_t), in most platforms it will work (in Windows it does).

When using std::string, does it have to be initialized? Is the following
code OK?

It is not required if I understand your question correctly.

string s;
S = "This is a test string";

s="This is a test string";



and for wstring:


wstring s=L"This is a test string";



L signifies a wchar_t literal.

Is it OK to have a container (for example, list), as global object whose
size and elements increases at run time?

Yes, but why make it global?






Regards,

Ioannis Vranos

http://www23.brinkster.com/noicys

 

[Default User]

Well, yes and no. (Sorry, Victor, I seem to be in the mood for quibbling
with you today <g>). A std::basic_string<wchar_t> works fine for Unicode
if wchar_t is large enough to hold individual Unicode characters. In
that case you only get zeroes if you put them there. If a wchar_t isn't
large enough you typically use one of the UTF-x encodings. The easiest
one to use portably is UTF-8, becuase you can always hold 8-bit
characters in a std::basic_string<char> (also known as std::string). A
UTF-8 encoded Unicode string only has zero values when you put them
there; the encoding doesn't generate spurious zeroes.

I've never used wide strings of any sort. If you have a wstring, what
does the call to c_str() do in the case where the string uses some
encoding like you mention. Does it convert each wide character into some
equivalent that a char can hold?




Brian Rodenborn

 

[Pete Becker]

I've never used wide strings of any sort. If you have a wstring, what
does the call to c_str() do in the case where the string uses some
encoding like you mention. Does it convert each wide character into some
equivalent that a char can hold?

basic_string<wchar_t>::c_str() returns a const wchar_t* that points to a
null-terminated sequence of wchar_t. No conversions.

 

[Default User]

basic_string<wchar_t>::c_str() returns a const wchar_t* that points to a
null-terminated sequence of wchar_t. No conversions.

Gotcha, thanks. Just curiosity on my part, as I said I've never used
wide chars.




Brian Rodenborn