auto_ptrimpl())
pain.

ok man, thank you, i think you only missing the -o helloworld in the end

yeah

the best way to do this maybe is using a Makefile?

pkakkar_, no I am not interessted sorry, I am too busy
mad_lord, sure

okay!

pkakkar_, I might help on a concrete problem
pkakkar_, but I have no time to work on another project

evilissimo, Can i have your email addr..

pkakkar_, if I am available for questions I will be here
but

so, this is the last question, i promess, do you know how can I insert the `pkg-config gtkmm-2.4 –cflags` attribute in the Makefile ?

there you go, pkakkar_
mad_lord, let me even do one

evilissimo,Thanks

how to read or write classes to files??

The term is called serialization.

mad_lord, http://snip-a-lot.de/gtkmm_template.tar.bz2

can anyone thinknof a neater way to write this? http://rafb.net/p/vvMSAW14.html

One way is to have that code in a macro like #define MOM_LOOK_THIS_IS_NEAT

i really dont like macros :/
theres no other way?

wow! thank you som much man

np yw

Dunno, it's too hot to think of neat code
Unless it's lisp

where are you?

Finland

it sure as shit aint hot here (UK)
what i really want is a way to clear a function object

when does template argument deduction doesnt work for functions? I know a case where the function argument type depends on the template argument (being a nested type of a template class type that depends on the template argument)
here are now arround 35 degrees officially tho on the street in the city I bet there are more than 45

d1zzy, the standard tells you, I think there was a clear clause that says when inferring doesn't work
See above regarding heat
And code

indeed
ok I'll look it up, didn't knew exactly where in the standard to look or what to search for

The standard term for it is "inferring", for easier grepping

ah found Template argument deduction, page 285

Oh, it isn't?

orly?

how long have you been doing ISO C++?

Since 1998 or so

heh

Adrinael never did c++, he's a lisp guy, so he's infinitely intelligent, which explains why he's active in 40 channels simultaneously

Hey, wait

Alternatively, he doesn't have a life.

Activity in window 59

recusion does that.

But who does?

does a job count as life ?

No

damn.

Does not having a job count as a life?

No!

There goes my options.

you get a life when you stop asking these questions on IRC

Will I stop asking questions?

I just had a life!

!8ball Will blank ever stop asking questions?

Maybe…

We have hope!

job or no job, just skeedaddle along, whistling a happy tune. That is life.

I think that's being canadian, not having a life

I can't whistle!
But if you're not Canadian it'll be a life.

So much Canada

/msg jkp no idea if it works like this but you might try: http://rafb.net/p/bqc0o917.html if you like it

blame Canada

haha

What a dirty private message!

:p

ta!

helping people priately, shame on you

evil villainHAHAHAHAHA/

Everytime you use boost, god masturbates.

nifty

poor god

must be nackered

that would be nice, but i think for it to be useful or worthewhile the ConstructDelegate template needs to be more geeneral

literally :p

thanks for the idea tho

jkp http://rafb.net/p/c2bOcE23.html

this is what i was thinking

That is starting to look more and more like bind.

you could have variants for the number of parameters too

Zao,

yes, but the problem with bind is that it doesnt detect or allow NULL

I'd say it's a flaw if you bind an member function to a null pointer.

can anyone tell me how to read and write classes to files??

!fs seria

serialization is http://www.parashift.com/c++-faq-lite/serialization.html | http://www.boost.org/libs/serialization/doc/index.html

abhee, someone already responded

when you declare a functor it can be used in a boolean experessioon to tell if it is set

abhee, it's called serialization.

You've already gotten the magic keyword "serialization".

!give abhee serialization

serialization is http://www.parashift.com/c++-faq-lite/serialization.html | http://www.boost.org/libs/serialization/doc/index.html

i hd got disconnected

hehe

so what i want is a way to say, if i give you a null pointer, give me that initiali functor back

Hello!

lo

look at this code, coped from Thinking in c++ book :
http://rafb.net/p/H2bPhS21.html
f5() = X(1);
what does it mean?

rthriller, what does line 14 do.

it returns an X object

So you have an instance of X on the left hand side.
What does X(1) do?

but it is unusal (at least for a beginner) to use a function on the left
but then the object created is lost?

After line 29, yeah.

so that line is useless
and line 30 doesn't use the object created in line 29

Well, the whole code is useless.
Right, line 30 is a new instance of X.

Yes it is.
so it is not common to put the func at the left side?

no

ok, thank you. That was really confusing

anyway, referring to "static" keyword, isn't it a bad situation that the same keyword has 3 meanings?

apparently Bjarne doesn't think so

hi all

Haha. static is hardly unique in that regard

anyone know what is the best* version of tic++'s pdf ?

Probably the non-corrupt version.

Arafangion, lol

since i got 4/5 version of pdfs. each of them has some advantage

ok but I ask: 2 meaning is already a bad situation in C… but 3 meanings in C++ is even worse!!!!!!

Consider '&' and '*'.
In some way it simplifies the syntax once you know it - Human kanguages require far more context.

no, it would be bad if they were confusing somewhat
but they are not confusing
they are actually very easy to learn and share common semantics
there are common features to all "static" usage out there
they dont just pick up some random feature and say let's use static host for it lol

i see

you know, they could have used "static" instead of "virtual" for virtual inheritance too

yeah

Metabol, now that would be confusing

but d1zzy why c++ introduced "private " keyword if it's right what you are saying?
but d1zzy why c++ introduced "private " keyword if it's right what you are saying?

What's virtual inheritance?

because they found no other keyword that shared common semantics with what private jvm web hosting should have done

class D : public virtual A { };

that's the point. if I am to decide how to implement per class members, obviously I think that "static" is a good candidate

Arafangion, if you multiply derive from two classes, and those two classes derive from the same base class, the virtual keyword makes that original base class singular

that's because the other statics too exist as long as the program exist and so on
same thing with "auto" for C++0x used for type inference
I think it's a good choice
what should they used instead?

i see

"decltype" on the other hand, meh

maybe the should have eliminated static (in the meaning of visibility)… but in this way it wouldn't had compatible with C
sorry for the horrible sentence
let's reorder it:

I think there are really other language problems in C++ than overloading static/auto
those 2 are really non-issues

which ones are them? can you tell me one of them as an example?

current C++ misses proper way to do move semantics, misses proper way to do perfect forwarding, misses a delete expression with additional parameters (as you can call new with additional parameters)

yeah, if you don't like it, see it as an endearing excentric quirk

those I am more sad about
with Mojo & co move semantics can be implemented but it's cumbersome and has dirty syntax
perfect forwarding it's very nice to have it to have wrappers for new to make all your objects return auto_ptr and such (new_auto_ptrType(args))
having additional arguments to delete as you can to new allows for a symmetric and scalable way to play with various memory pools using new/delete expressions
but in the end none of those are showstoppers
being a C programmer for a long time I really can't complain about C++'s lack of functionality (that would be absurd heh)

id like a better way of implementing forwarding i have to say

i see
thnks

C++ has some ugly things from C too, array-pointer degradation, lame switch()

so far I consider it the perfect blend of low-level coding functionality with highlevel (even compile time metaprogramming) coding. and because my interests (and job) are arround system programming, C++ fits very nicely there

What's wrong with the switch? (Though it'd be neat if one could allow more complex types in switches)

yes, more complex types / full blown expressions, and no automatic drop-through between cases

Like in Common Lisp!

i know that this is off topic but do you know any good guide for low-level programming in C ?

I don't see why C++ couldn't have such features, except for the drop-through thing. (Which I personally like)

right now it just looks like syntactic sugar for an assembly jump table

Which is what it is.

yes!
But I don't lose sleep over it because switch() does not conform to the OO paradigm anyway

I disagree.
Though… In OO code, many people /do/ misuse it.
For example, in my code, I have: keyHit(key) { switch(key) { … }}

that was kind of a tongue in cheek remark about the OO paradigm there

Many of the design patterns in Fowler's book seem to be focused on how to abuse switch cases and how to therefore "improve" the code.

Analysis Patterns? Haven't read it
So many Fowler books

To be honest, I haven't read it either, but I'll bet that they abuse switches

I do lose some sleep over 1) the lack of, whatchmacall it, compiletime and/or runtime introspection, I don't know the term. For things like serialization, database binding, widget binding. But I don't know enough to see how it should work
2) the fact that C++ is hard to parse

(2) is quite true.
There are alternative syntaxes, though, afaik.

Does there even exist a "refactoring" tool which does something so simple as a safe project-wide renaming of a member variable? Is it even possible to write?

vs does that, iirc.

Borland Developer Studio

You'd need a miracle to catch macro appends and the like.

SlickEdit

yes

fwiw, 1) is what we try to solve with the moc tool in Qt

Or you could write a parser and Do It Right

ast, change, convert ast back to files pre-tu, work out what's changed — not trivial)

tu ?

!tu

A source file together with all included headers and source files, less any source lines skipped by preprocessor conditionals, is called a translation unit, or TU. TU's can be separately translated and then later linked to produce an executable program.

ah

FauxFaux, I think it needs to be coupled with the actual compiler one uses, then, barring preprocessor tricks, it's possible

Yeah. I know for a fact that VS' parser and compile hosting are different (yeah, bitch away), so their's wouldn't be foolproof.

bhughes, thanks, that is worth something, because, and you have probably heard this before, the principle of the moc can be intimidating

all the time, yes

I can think of some legitimate cases where you'd need to cope with simple preprocessor conditionals (state pointers being different on different platforms etc.)

i have to agree with you though, i would love to see meta information as part of the language

Do you know why switch-case only works with integers?

switch-case will work with any constant expression that can be evaluated at compile time

you probably mean runtime.

no, i meant compile time

no, compile hosting time

for the case blocks, that is

is this also in the standard or compiler-specific ?

is not allowed

kirkt, it's standard

the expression in switch() can be evaluated at runtime

Metabol can u refer me to how to find it there ?

is 'function polymorphism' used a lot?

kirkt, standard in my head is not, sorry

My theory is that because ints can be represented as 1 \/ 2 \/ 3 \/ … \/ 2^32, as an discriminated union of numbers, thus they can be used in switch-case.

ok found it

heh?

\/?

\/ == or.

gotcha

http://rafb.net/p/YFUiza63.html and I am still getting aborted by SIGPIPE :/

strange… which version of Qt?

!rq

it might be shunned by the religion of C++.

4

(be prepared, we may get chased off to #qt ;D)

:p
this is an issue with signal handling

are you using pipes in your code?
threads?

!rq

no sockets

i am merely a padawan amongst masters

!rq

What a helpful channel you guys are

I get SIGPIPE while trying to write to a disconnected socket I guess

yup

just want to ignore SIGPIPE

no threads?

qt opens a new thread?

http://www.opengroup.org/onlinepubs/000095399/toc.htm

if yes then the answer is yes

no, i'm asking if you are using threads

no I am not

k
http://www.opengroup.org/onlinepubs/009695399/functions/sigaction.html
that's a good place to learn about signal handling
i have to say though, if you're using Qt 4, why not use QtNetwork?

does that handle OS signals too?

Hey. I'll jump right into asking questions… Hope that's ok…
I'm trying to do a stacktrace on win32 (MSVC), but I'm having problems. I get the first frame right, but then StackWalk64 returns false on the next… This only happens in optimized mode (but with debug info), but MSVC's internal debugger can properly show the callstack.
Anyone happen to know a lot about StackWalk64?

it eliminates the need to deal with them, yes

ok thanks I will take a look

http://doc.trolltech.com/4.3/qtnetwork.html

int s0 = someString.find_last_of(someChar); ?

std