C++ is becoming a freak language that's parading its disfigurements in front of mildly disgusted but curiously fascinated audience.

Unfortunately, there isn't a clear successor to C++ lined up. Java and C# had their chance and it became clear that, while they simplify some aspects of development, they don't put us on the wholly new effort/reward curve the industry needs to be on.

int array[100];
int *p=...; // указывает на элемент array

size_t index=p-array; // в это строчке -- неявное преобразование

Talk to me dirty! Tell me more about name resolution!

Безусловно, Java и C# намного элегантнее, но чтобы их
придумать, надо было пройти тернистый путь C++ , набить все шишки и
получить бесценный опыт.

class A
 {
 };

class B : public A
 {
 }


B * pb = new B;

A * pa = pb; // пытаются заменить на A * pa = (A *)pb;, что ИМХО никому не нужно

delete pb; // а то сейчас опять про меморилик вспомнят :)

I talked to John Lykos at one of the conferences, and he told me that he would pay big bucks for a tool that would be able to tell what header files must be included in a given file.

class A
 {
 };

class B : public A
 {
 }


B * pb = new B;

A * pa = pb; // пытаются заменить на A * pa = (A *)pb;, что ИМХО никому не нужно

delete pb; // а то сейчас опять про меморилик вспомнят :)

Basic Whitespace Overloading

Here, I describe the more innovative and powerful mechanism for overloading whitespace. Consider x*y.
In programming languages (e.g. Fortran, Pascal, and C++), this is the conventional notation for
multiplying two values. However, mathematicians and physicists traditionally do not use the operator *.
Instead they use simple juxtaposition to indicate multiplication. That is, for variables x and y of suitable
types,

x y

means multiply x by y.
This is simply achieved by overloading the space operator for double-precision floating-point values:

double operator (double d1, double d2) { return d1*d2; }

Or — more explicitly — equivalently

double operator ' '(double d1, double d2) { return d1*d2; }

Given one of these definitions, a physicist can use his (or her) conventional notation rather than the
notation that has become conventional among computer scientists:

double f(double x, double y, double z)
{
return a + x y pow z; // using also a user-defined operator pow
}

Clearly, the space operator has (by default) a precedence lower than pow and higher than +. The
mechanism for assigning precedence to user-defined operators is described in detail in the companion
article.

Д>class A
Д> {
Д> };

Д>class B : public A
Д> {
Д> }


Д>B * pb = new B;

Д>A * pa = pb; // пытаются заменить на A * pa = (A *)pb;, что ИМХО никому не нужно

Д>delete pb; // а то сейчас опять про меморилик вспомнят :)
Д>

Д>Wish you happy debugging

Д>Только не надо кричать "да я! да я никогда!" и бить себя пяткой в грудь. Все когда-нибудь ошибаются...

shared_ptr< A > a( new B() );

User-visible-changes:
— Client:
* fixed: 'svn switch' interruption can break working copy (issue #1826)
* fixed: 'svn switch' memleak over ra_dav (issue #2106)
* fixed: 'svn blame' algorithm bug (r11527)
* fixed: invoke external diff/diff3 with local-style paths (r11689)
* fixed: 'svn status' handling of missing subdirs (r11936)
* fixed: 'svn ls -v' encoding bug (r11740)
* fixed: 'svn ls "file with space"' bug (r12273, r12393)
* fixed: 'svn merge' should URI-encode copyfrom URLs (issue #1905)
* fixed: 'svn merge' deletion output formatting (r12100, r12111, r12114)
* fixed: 'svnversion --version .' crash (r11438)
* fixed: UNC paths on Cygwin (issue #2108)
* fixed: win98 iconv bug -- uninitialized variable (issue #2091)
* improved 'svn status' performance:
— do fewer check_path calls (r11592)
— 'svn status file' shouldn't recursively lock tree (r11439, r11669)
* translation updates for localized client messages.

— Server:
* fixed: 'svnadmin load' race condition (r12327)
* fixed: fsfs memleak in commit finalization (r11706)
* fixed: fsfs memleak in inefficient directory removal (r11701)
* fixed: fsfs commits use insert-only perms on db/revs/ (r11665)
* fixed: fsfs creates lockfile at creation time, not at 1st commit (r12172)
* fixed: svndumpfilter mislabeling output as version 3 (issue #2142)
* fixed: 'svnserve -h' encoding bug (part of issue #1997)
* fixed: prevent cross-repository copies (r12003)
* fixed: increase log-region max size in default DB_CONFIG (issue #2159)

— Both:
* fixed: 'svn switch' quietly corrupting working copy (issue #2124)
* fixed: canonicalize paths sent by ra_svn/svnserve (issue #2119)
* fixed: memleak into UTF8 translation routines (r11689)

Developer-visible-changes:
* add support for BerkeleyDB 4.3
* add support for any apr/apr-util 1.X
* disallow incompatible SWIG versions (r12450)
* fixed: slight API/ABI incompatibility between 1.0.9 and 1.1.x (r12102)
* fixed: perl bindings pool usage & object refcounts (r11451, r11630)
* fixed: perl bindings pool usage and potential memleak (r12397)
* fixed: javahl crash trying to fetch nonexistent property (r12184)
* fixed: javahl build can fail due to missing dirs (issue #2032)
* fixed: RPM build breakage (issue #2111)
* fixed: i18n issues for windows installer (r11685)
* allow build system to update single .po file (r11763)

   struct bidirectional_iterator_tag: public forward_iterator_tag {};
   struct random_access_iterator_tag: public bidirectional_iterator_tag {};

template <class Arg1, class Arg2, class Result>
struct binary_function {
   typedef Arg1 first_argument_type;
   typedef Arg2 second_argument_type;
   typedef Result result_type;
};

template <class T> struct plus : binary_function<T,T,T> {
   T operator()(const T& x, const T& y) const;
};