| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9
Comments:
<0> eg, 4-byte boundries.
<1> it's a struct
<1> and i am moving it futher into an array
<0> well, the elements of the struct then.
<1> by an arbitrary ammount
<0> pretty much everything except for 'char' requires alignment
<1> still, even if it needs to be aligned
<0> so directly incrementing only by a byte is not valid, in general. so, you will need to do it the long way.
<1> my struct is greater than sizeof(long)
<0> yeah, but this is a very rare operation, and so it is not something you ordinarily need to do.
<1> ok, so there is no other way apart from casting to char and back
<1> in three c statements
<0> correct
<1> ok, that's a shame
<0> if you are directly working with memory, you should expect to work on char *
<2> use a void *
<0> pointer arithmetic is invalid on void*
<0> char* is the proper base type for memory manipulation.
<1> well i've actually managed to avoid char* until now
<0> dila, but where did the buffer come from in the first place?
<1> it's a heap
<0> ah, well, yeah, it's expected that low-level memory operations, such as this one, will be using char *
<1> hmm
<1> but you see i have a struct that has the size of the bufer
<1> and i'm moving the buffer further into the heap
<0> off the end of the buffer?
<1> so i don't actually need to use char*
<1> never mind
<1> the problem is with c syntax
<1> the code is fine
<0> hehe
<1> oh well
<1> that was the last line of my heap cl***
<1> which is the only part i am anoyed about
<1> since it's three c statements for an operation that is a single ***embly instruction
<1> IA32, anyway
<0> most heap implementations work on char *, and only cast to struct (for bookkeeping info, etc) as necessary.
<0> however, i still suspect you are ignoring alignment considerations, which will not work on many non-ia32 architectures.
<0> you're free to ignore alignment on x86, but there is a performance penalty for it.
<0> dwords, for eg, should be 4-byte aligned.
<2> saying your chosen alignment is chaotic evil has no meaning here
<1> well the new[] operator can enforce alignment if necessary
<0> im not sure it can, but thats not what i was talking about.. i was talking about the alignment of the members of your bookkeeping struct.
<0> and as far as i know, all returns from operator new need to have 'universal alignment'
<1> yes
<1> so if new[] rounds the size of the requested buffer to a four byte boundry it will guarentee that the bookkeeping structs are also aligned
<0> ok, i guess. i don't really understand what you're doing.
<1> it's just that the heap cl*** is capable of performing un-aligned allocs
<1> but the heap cl*** isn't going to be the primary interface for memory allocation
<1> the heap cl*** doesn't work anyway
<1> i have descoveredd
<1> which i guess you might find amusing :)
<1> oh well, thanks for the help
<3> gcc allows pointer arith on void*. This is nonstandard expention.
<2> it is a bit bizzare too , because void is nothing so can't have a length
<3> yeah
<2> does it work the same as pointer arithmetic on a char *
<3> yeas
<0> http://www.college***advice.com/*********ion-mistakes.shtml
<4> lol...
<4> .
<4> lol
<3> AaronWL: did you sleep today ?
<2> AaronWL, I like number 17 :)
<5> Hi all. I appologize, I'm not a C++ guy but Im trying to use SWIG to generate a wrapper for a simple header file. Anyone done anything like this?
<6> why not ask in #c++ ?
<7> pwnd
<5> Because it said help questions should be directed here
<6> because they're lying ****s
<5> Those bastards :)
<8> haha
<8> he asked in #C++
<8> let me just tell him to ask in #C...
<5> I'm more then willing to just pay someone to do this, if anyone is interested.
<7> lol
<7> how much?
<7> $120 an hr
<5> Take a look at the header file and let me know what you think it's worth.
<7> 3hr minimum
<5> And that would be more then its worth.
<7> sorry, my time is valuable
<5> Im sure, but this project isnt :)
<7> ask a n00b then
<6> If I knew wtf you were talking about it'd do it for $50
<5> zid_: http://rafb.net/paste/results/ZT6YWG92.html
<5> Thats the header file. I need the functions defined by that file to be accessible to either .NET or Java.
<5> doh :(
<5> And I know .NET and Java, but no C++ :) Certainly not enough to do this simple task.
<6> what's the problem anyway, aren't the names getting mangled how .net wants them?
<9> Java can call functions with C binding but not C++ binding as far as I know.
<5> Its the use of callbacks, hwnds and structs that are screwing me up.
<6> oh so everything exports ok?
<5> I can hack together this type of thing with simple code, but all the custom crap in that header file is killing me because I dont know C++
<5> java can natively call C++
<9> You could, of course, write C functions wrappers for the C++ functions. Then, Java could call the C functions.
<9> It can? That's news to me.
<5> yeah, let me find the link on Sun's site.
<10> java can call "native functions"
<5> This will do:
<5> http://www.javaworld.com/javaworld/javatips/jw-javatip17.html
<11> how can i return a char array from a function?
<10> zerocide: you must either malloc a char array that you return a pointer to or the preferred method would be to p*** a pointer to an array to the function that the function would fill
<11> ah
<11> yeah, i did the 'preferred' method
<11> was just curious about actually returning it
<11> thanks :)
<10> a more-creative method would be to create a struct that contains an array and then return the struct
<6> OrngeTide?
<11> oh
<11> that sounds nice
<12> Returning the struct is very inefficient, return or p*** a pointer
<10> Maloeran: it's all relative ;-) most c++ programmers use a lot of inefficient copies
<11> why is it inefficient?
<11> does it return a copy of the struct?
<10> yes
<11> oh, heh
<11> but it is possible to return a pointer to it
<11> isn't it?
<10> you can't return a pointer to a local variable/struct/array. you'd have to malloc the struct
<12> If it was malloc'ed, sure, or make the calling function hands a pointer to the struct
<10> which you're not gaining anything over using an array
<11> oh
<11> oh yeah, i can see why
<13> Arses
<14> (: hi
<13> ****er
<13> I mean erm, hi
<15> hi
Return to
#c
or
Go to some related
logs:
disable connection tab
suna thermometer
#stocks
#hardware
#vmware
hardware-ram-tester
hooking Createprocess
donkey dink
#sex
#firebird
|
|