class string

[cmk128]

Hi
I am going to implement a new string class for my os, do you have
any idea to enhance of standard c++ string class?
If compare c++ string class with CS++ CString class and java String
class, what is your opinion?
thanks
from Peter (cm****@hotmail.com)

[Alipha]

cm****@hotmail.com wrote:

Hi
I am going to implement a new string class for my os, do you have
any idea to enhance of standard c++ string class?
If compare c++ string class with CS++ CString class and java String
class, what is your opinion?
thanks
from Peter (cm****@hotmail.com)

we don't need yet another string class! Use the standard string class.
I'm tired of all the bullshit of having to convert among CString,
wxString, qtString, std::string, char*, etc. Just stick with
std::string. If you want to add functionality, put free-standing
functions in a header file and include that.

[Alexei A. Frounze]

"Alipha" <al*****@hotmail.com> wrote in message
news:11*********************@o13g2000cwo.googlegro ups.com...

cm****@hotmail.com wrote:

Hi
I am going to implement a new string class for my os, do you have
any idea to enhance of standard c++ string class?
If compare c++ string class with CS++ CString class and java String
class, what is your opinion?
thanks
from Peter (cm****@hotmail.com)

we don't need yet another string class! Use the standard string class.
I'm tired of all the bullshit of having to convert among CString,
wxString, qtString, std::string, char*, etc. Just stick with
std::string. If you want to add functionality, put free-standing
functions in a header file and include that.

Shouldn't Unicode text need special string handling functions? :) That'll be
another class.

Alex

[Ivan Vecerina]

<cm****@hotmail.com> wrote in message
news:11**********************@o13g2000cwo.googlegr oups.com...

I am going to implement a new string class for my os, do you have
any idea to enhance of standard c++ string class?
If compare c++ string class with CS++ CString class and java String
class, what is your opinion?
thanks

Same advice as Alipha: best is to just use std::string
(assuming a complete C++ toolchain exists for "your os").
If you see a reason not to do so, please discuss it here.
Strings are an over-researched topic. The string class
defined by the C++ Standard only has two weaknesses IMO,
and they probably aren't what you would think:

1) std::string has too many member functions
There is no end to the number of functions/transforms one
may want to apply to std::string. The right way to deal
with the situation is to implement only essential operations
as member functions, and all the rest as 'free' functions.
Most C++ experts agree that too much has already been
included as members of std::string.
(yes, other string classes have done worse...)

2) there could be a place for a non-mutable string
If the characters of an std::string instance were not
modifiable (as is the case in Java), some optimizations
could be applied. Some library writers question whether
having only a non-mutable string, or both std::const_string
and std::string, would be better that the current situation.
hth -Ivan
--
http://ivan.vecerina.com/contact/?subject=NG_POST <- email contact form

[sap.liu]

if you handle unicode text string, you should use class template of
basic_string<>, which can be adaptable for every special multi-bytes
string.

[Ivan Vecerina]

"Alexei A. Frounze" <al*****@chat.ru> wrote in message
news:3n***********@individual.net...

"Alipha" <al*****@hotmail.com> wrote in message
news:11*********************@o13g2000cwo.googlegro ups.com...

cm****@hotmail.com wrote:

> Hi
> I am going to implement a new string class for my os, do you have
> any idea to enhance of standard c++ string class?
> If compare c++ string class with CS++ CString class and java String
> class, what is your opinion?
> thanks
> from Peter (cm****@hotmail.com)

we don't need yet another string class! Use the standard string class.
I'm tired of all the bullshit of having to convert among CString,
wxString, qtString, std::string, char*, etc. Just stick with
std::string. If you want to add functionality, put free-standing
functions in a header file and include that.

Shouldn't Unicode text need special string handling functions? :) That'll
be
another class.

Actually not:
- Unicode text encoded as UTF-8 can be used with std::string as is.
Your platform's default wide-character (unicode) encoding
is probably supported by std::wstring already.
Other unicode encodings would be implemented with an other
character and character trait parameter to std::basic_string
(on which the std::string typedef is based).
- Any Unicode-specific operations would best be implemented
as non-member functions.
Regards, Ivan
--
http://ivan.vecerina.com/contact/?subject=NG_POST <- email contact form

[Alexei A. Frounze]

"Ivan Vecerina" <IN*************************@vecerina.com> wrote in message
news:de**********@news.hispeed.ch...

"Alexei A. Frounze" <al*****@chat.ru> wrote in message

....

Shouldn't Unicode text need special string handling functions? :) That'll be
another class.

Actually not:
- Unicode text encoded as UTF-8 can be used with std::string as is.
Your platform's default wide-character (unicode) encoding
is probably supported by std::wstring already.
Other unicode encodings would be implemented with an other
character and character trait parameter to std::basic_string
(on which the std::string typedef is based).
- Any Unicode-specific operations would best be implemented
as non-member functions.

Usually, if there's any text processing other than just copying the entire
text or appending text, the functions must know the internal structure of
the text and Unicode is something very rich in this aspect (consider not
just the variable length of representation of a single code point but also
the combining marks and other things). So, is there any good and well-known
implementation of text/string handling for Unicode, compliant with the
standard?

Alex

[j]

"Alexei A. Frounze" <al*****@chat.ru> wrote in message
news:3n***********@individual.net...

"Ivan Vecerina" <IN*************************@vecerina.com> wrote in message news:de**********@news.hispeed.ch...

"Alexei A. Frounze" <al*****@chat.ru> wrote in message ...

Shouldn't Unicode text need special string handling functions? :) That'll be
another class.

Actually not:
- Unicode text encoded as UTF-8 can be used with std::string as is.
Your platform's default wide-character (unicode) encoding
is probably supported by std::wstring already.
Other unicode encodings would be implemented with an other
character and character trait parameter to std::basic_string
(on which the std::string typedef is based).
- Any Unicode-specific operations would best be implemented
as non-member functions.

Usually, if there's any text processing other than just copying the entire
text or appending text, the functions must know the internal structure of
the text and Unicode is something very rich in this aspect (consider not
just the variable length of representation of a single code point but also
the combining marks and other things). So, is there any good and

well-known implementation of text/string handling for Unicode, compliant with the
standard?

Alex

wstring is for utf16 or utf32 afaik..
utf8 is not generally supported...

[Ivan Vecerina]

"Alexei A. Frounze" <al*****@chat.ru> wrote in message
news:3n***********@individual.net...

"Ivan Vecerina" <IN*************************@vecerina.com> wrote in
message
news:de**********@news.hispeed.ch...

"Alexei A. Frounze" <al*****@chat.ru> wrote in message

...

> Shouldn't Unicode text need special string handling functions? :) That'll > be
> another class.

Actually not:
- Unicode text encoded as UTF-8 can be used with std::string as is.
Your platform's default wide-character (unicode) encoding
is probably supported by std::wstring already.
Other unicode encodings would be implemented with an other
character and character trait parameter to std::basic_string
(on which the std::string typedef is based).
- Any Unicode-specific operations would best be implemented
as non-member functions.

Usually, if there's any text processing other than just copying the entire
text or appending text, the functions must know the internal structure of
the text and Unicode is something very rich in this aspect (consider not
just the variable length of representation of a single code point but also
the combining marks and other things). So, is there any good and
well-known
implementation of text/string handling for Unicode, compliant with the
standard?

std::string does not have built-in support for all possible operations on
Unicode text, or for path-manipulation operations, or for loading inter-
nationalized text from resources, or whatever processing you have in mind.

But these operations can be implemented as non-member functions, and this
is the right way to support them.

Think of C++ containers and its library of standard algorithms, which
are independent and orthogonal. A much better, lean and efficient design,
than developing an ultra-container trying to do everything.
Cheers, Ivan
--
http://ivan.vecerina.com/contact/?subject=NG_POST <- email contact form
Brainbench MVP for C++ <> http://www.brainbench.com

[Alexei A. Frounze]

"j" <je****************@stofanet.dk> wrote in message
news:43**********************@nntp02.dk.telia.net. ..
....

wstring is for utf16 or utf32 afaik..
utf8 is not generally supported...

Most likely utf-16, easier than utf-8 and less space waste than utf-32.
But anyway, I presume there's not much (if any) support for what Unicode is.

Alex

[Alexei A. Frounze]

"Ivan Vecerina" <IN*************************@vecerina.com> wrote in message
....

std::string does not have built-in support for all possible operations on
Unicode text, or for path-manipulation operations, or for loading inter-
nationalized text from resources, or whatever processing you have in mind.

But these operations can be implemented as non-member functions, and this
is the right way to support them.

Think of C++ containers and its library of standard algorithms, which
are independent and orthogonal. A much better, lean and efficient design,
than developing an ultra-container trying to do everything.

That was the sad point. It still needs to be implemented since apart reading
and writing code points everything else is quite a big task.

Alex