- "Linus Torvalds Was (Sorta) Wrong About C++" - 23 Updates
- Need Help with design of class hierarchy - 1 Update
- Cgi and c++ - 1 Update
| Christian Gollwitzer <auriocus@gmx.de>: Mar 11 08:11PM +0100 Am 11.03.15 um 19:46 schrieb JiiPee: > array<int> a(10); > Which take objects allocation (couple of bytes) plus 10. So C makes > shorter executable what many want. Are you sure this is correct? My understanding of std::array vs. std::vector is that the latter gives you dynamic memory with it's benefits (dyanamic growth etc.) and drawbacks (overhead), while the former gives you zero overhead compared to a C-style overhead but static type safety. Christian |
| JiiPee <no@notvalid.com>: Mar 11 07:14PM On 11/03/2015 18:56, Mr Flibble wrote: > "std::array<char> a;". A std::array does not perform any allocations; > it's storage is inline just like for a C style array. > /Flibble ok sorry, might be.... then it was: std::vector<int> which takes extra allocation for the object. And now.. dont say this is not true as I have definitely tested it myself!!! :) and in real life we really most of the time need that vector-system. So dynamical C-array is smaller |
| JiiPee <no@notvalid.com>: Mar 11 07:21PM On 11/03/2015 19:11, Christian Gollwitzer wrote: > former gives you zero overhead compared to a C-style overhead but static > type safety. > Christian okey, array maybe not, but vector<int> for sure. there is an object overhead |
| Christopher Pisz <nospam@notanaddress.com>: Mar 11 02:31PM -0500 On 3/11/2015 2:14 PM, JiiPee wrote: > ok sorry, might be.... then it was: > std::vector<int> which takes extra allocation for the object. And now.. > dont say this is not true as I have definitely tested it myself!!! :) did you use std::vector::reserve in your test? > and in real life we really most of the time need that vector-system. So > dynamical C-array is smaller char a[10]; is not a _dynamic_ array. -- I have chosen to troll filter/ignore all subthreads containing the words: "Rick C. Hodgins", "Flibble", and "Islam" So, I won't be able to see or respond to any such messages --- |
| JiiPee <no@notvalid.com>: Mar 11 07:32PM On 11/03/2015 18:57, Mr Flibble wrote: >> it's storage is inline just like for a C style array. > Er, I of course meant "std::array<char, 10> a;". > /Flibble Flibble.... i just faced this issue recently when somebody asked me to change their C-code to C++. Its a music hardware which has only like 64K memory. He wanted to save every byte. I first said "well, I would use std::vector.... lets install std stuff". But later I started to agree with him that plain old C-array takes less space and he does not need really that vector there (well, was it vector or array i dont remember, but anyway.. the idea that objects take more space). Although now as am thinking, he created many other classes and objects willingly, surely they also take extra memory... so its not so obvious which is best. |
| Christopher Pisz <nospam@notanaddress.com>: Mar 11 02:33PM -0500 On 3/11/2015 12:29 PM, Lynn McGuire wrote: > famous naysayers and their profane soliloquies about inelegance dissuade > you." > Lynn Who cares what Linus Torvalds says about anything other than the Linux Kernel? -- I have chosen to troll filter/ignore all subthreads containing the words: "Rick C. Hodgins", "Flibble", and "Islam" So, I won't be able to see or respond to any such messages --- |
| Paavo Helde <myfirstname@osa.pri.ee>: Mar 11 02:35PM -0500 > dont say this is not true as I have definitely tested it myself!!! :) > and in real life we really most of the time need that vector-system. So > dynamical C-array is smaller You need std::vector in real life because you want dynamic resizing or maybe because the array is just too large to allocate on stack. In both cases this can be done in C as well, but then you need an additional pointer and maybe also another variable to hold the array size - which again produces about the same overhead than std::vector! In the extreme case where each byte must be conserved, I agree that with C- style arrays and lots of work one can win some bytes. But as C is a proper subset of C++, the same code can then be used as C++ code and still have better type-checking and other goodies. hth Paavo |
| JiiPee <no@notvalid.com>: Mar 11 07:37PM On 11/03/2015 19:35, Paavo Helde wrote: > better type-checking and other goodies. > hth > Paavo yes, and thats what I did with one of my clients... we did a lot of C code, but added type checking with C++ style as much as possible (or at least I recommend them) |
| JiiPee <no@notvalid.com>: Mar 11 07:38PM On 11/03/2015 19:35, Paavo Helde wrote: > better type-checking and other goodies. > hth > Paavo like they were using #define MY_VARIABLE 6 and i said "this really must be: const int MY_VARIABLE = 6; " he first hesitated because he was worried it takes more memory, but he googled and found that not....so he accepted :) |
| Ian Collins <ian-news@hotmail.com>: Mar 12 08:39AM +1300 JiiPee wrote: >> former gives you zero overhead compared to a C-style overhead but static >> type safety. > okey, array maybe not, but vector<int> for sure. there is an object overhead No more or less than the overhead for a dynamic array in C. -- Ian Collins |
| Mr Flibble <flibbleREMOVETHISBIT@i42.co.uk>: Mar 11 07:50PM On 11/03/2015 19:21, JiiPee wrote: >> Christian > okey, array maybe not, but vector<int> for sure. there is an object > overhead The "object overhead" of std::vector<int> is constant (O(1)) so becomes insignificant as n increases; the space complexity of std::vector, dynamic C arrays and static C arrays is identical: O(n). If you have lots of vectors of small n you are probably doing it wrong. /Flibble |
| JiiPee <no@notvalid.com>: Mar 11 07:54PM On 11/03/2015 19:39, Ian Collins wrote: >> okey, array maybe not, but vector<int> for sure. there is an object >> overhead > No more or less than the overhead for a dynamic array in C. std::vector<int> a; std:: cout << sizeof(a) << std::endl; prints 12 int* b = nullptr; std:: cout << sizeof(b) << std::endl; prints 4 So an empty vector seems to take 12 bytes, and empty C-array 4 bytes |
| Mr Flibble <flibbleREMOVETHISBIT@i42.co.uk>: Mar 11 07:58PM On 11/03/2015 19:54, JiiPee wrote: > std:: cout << sizeof(b) << std::endl; > prints 4 > So an empty vector seems to take 12 bytes, and empty C-array 4 bytes to be equivalent you also need to record the size of the C-array which is an extra 4 bytes and also an extra 4 bytes if you want to support vector's capacity which makes 12 bytes. /Flibble |
| JiiPee <no@notvalid.com>: Mar 11 08:05PM On 11/03/2015 19:58, Mr Flibble wrote: > is an extra 4 bytes and also an extra 4 bytes if you want to support > vector's capacity which makes 12 bytes. > /Flibble if the size is the same for different types of variables, then only one variable would do the job instead of multiple. like: int size = 10; int* a = new int[size ]; float* b = new float[size ]; double* c = new double[size ]; so in C we need total : 4 bytes overhead in C++: vector<int> a = ...10); vector<float> b = ...10); vector<double> c = ..10).; would need total: 36 bytes of overhead |
| JiiPee <no@notvalid.com>: Mar 11 08:06PM On 11/03/2015 19:33, Christopher Pisz wrote: >> Lynn > Who cares what Linus Torvalds says about anything other than the Linux > Kernel? come on, be careful... he is from my country: Finland!!! :) |
| Wouter van Ooijen <wouter@voti.nl>: Mar 11 09:10PM +0100 JiiPee schreef op 11-Mar-15 om 8:32 PM: > change their C-code to C++. Its a music hardware which has only like 64K > memory. He wanted to save every byte. I first said "well, I would use > std::vector.... lets install std stuff". Why use a std::vector, which is a flexible array, if you want a fixed array? You can't blame the language (or rather, the library) for giving you options that are wrong for your situation. > Although now as am thinking, he created many other classes and objects > willingly, surely they also take extra memory... so its not so obvious > which is best. A class does not need to take more memory than the data it stores and the code that implemnts its methods. Wouter van Ooijen |
| Mr Flibble <flibbleREMOVETHISBIT@i42.co.uk>: Mar 11 08:21PM On 11/03/2015 20:05, JiiPee wrote: > vector<float> b = ...10); > vector<double> c = ..10).; > would need total: 36 bytes of overhead If you pass your dynamic C array "a" to a generic function how does that function know how many elements there are? You need to couple the pointer with a size for dynamic arrays (just like vector does) to do anything useful. If "size" is always fixed at 10 then you don't need dynamic arrays and you are not comparing like with like when comparing with vector as vectors can grow. Your examples and analysis of the situation is artificial and naive; read my other reply regarding the space complexity of the various options. /Flibble |
| JiiPee <no@notvalid.com>: Mar 11 08:25PM On 11/03/2015 20:21, Mr Flibble wrote: >> would need total: 36 bytes of overhead > If you pass your dynamic C array "a" to a generic function how does > that function know how many elements you pass the size-variable with it. (size is 10 here). so where ever you use a -array you pass size with it. |
| Greg Martin <greg.at.softsprocket@dot.com>: Mar 11 08:26PM >> Lynn > Who cares what Linus Torvalds says about anything other than the Linux > Kernel? A lot of people as it happens. Whether they should or not is another matter. -- http://www.softsprocket.com |
| Mr Flibble <flibbleREMOVETHISBIT@i42.co.uk>: Mar 11 08:31PM On 11/03/2015 20:25, JiiPee wrote: >> that function know how many elements > you pass the size-variable with it. (size is 10 here). so where ever you > use a -array you pass size with it. And passing it with it everywhere is equivalent to it being part of it .. so 8 bytes of overhead. From my other reply: the "object overhead" as you call it of std::vector is only significant for small n. If you have lots of vectors of small n then you are probably doing it wrong. /Flibble |
| Paavo Helde <myfirstname@osa.pri.ee>: Mar 11 04:15PM -0500 > int* a = new int[size ]; > float* b = new float[size ]; > double* c = new double[size ]; This is not exactly equivalent to std::vector because the capacity and efficient dynamic resizing are missing. > so in C we need total : 4 bytes overhead 16 bytes, if you want to compare correctly. Each pointer is an overhead. And if you add capacities, it will make 28 bytes. > vector<float> b = ...10); > vector<double> c = ..10).; > would need total: 36 bytes of overhead If the vectors are always of the same length, then the solution is clear: struct X { int a; float b; double c; }; std::vector<X> x; Voila: this has 12 bytes overhead, which is 4 bytes less than the C version, plus it supports efficient dynamic resizing as a bonus, plus it is not error-prone and exception-unsafe - an even bigger bonus. Q.E.D. Cheers Paavo |
| Mr Flibble <flibbleREMOVETHISBIT@i42.co.uk>: Mar 11 09:29PM On 11/03/2015 21:15, Paavo Helde wrote: > Voila: this has 12 bytes overhead, which is 4 bytes less than the C > version, plus it supports efficient dynamic resizing as a bonus, plus it > is not error-prone and exception-unsafe - an even bigger bonus. Q.E.D. You could of course have a dynamic array of X also. In my most humble opinion dynamic arrays should only be used for one thing: allocating uninitialised buffers of char; use std::vector for everything else.. /Flibble |
| JiiPee <no@notvalid.com>: Mar 11 09:31PM On 11/03/2015 20:31, Mr Flibble wrote: >> use a -array you pass size with it. > And passing it with it everywhere is equivalent to it being part of it > .. so 8 bytes of overhead. it remains 4 bytes like i originally calculated. size is a global variable (defined in a same place where "a" is defined). Why 8? size is 4 bytes.... you mean the function needs another size variable as an argument? |
| Victor Bazarov <v.bazarov@comcast.invalid>: Mar 11 03:36PM -0400 > class Song // every song has it's own class > string &name; // full path to song on disk. > virtual GetRelease(); // With a pointer to any Song class, one virtual WHAT GetRelease() ? What's the return type? > virtual GetFullPath(); / /Artist associated with that song. > class Disc : Song // Each Release has at least 1 disc > virtual int GetDiscNumber(); 'Disc' derives from 'Song'? So a 'Disc' IS-A 'Song'? I am not sure I actually agree. Shouldn't 'Disc' be actually a container of 'Songs'? > Looking for suggestions on how one would go about designed a good class > hierarchy to represent the music of a particular recording artist. > Any suggestion greatly appreciated. OK, here is a suggestion: start from the right end. What you need is a system that *uses* your objects. From their *use* you can distill their *interface*. Once you have the interface, the class design becomes a walk in the park. So, start with describing how your objects are going to be *used*. And it doesn't really matter what language you are going to use to implement your system. Don't try designing with any specific language in mind. Forget 'virtual' or 'array<>' or "pointer to". Use pseudo-code to describe the relationships. Translating your pseudo-code into any language of your choice is not important to the design, really. What if tomorrow you decide to make a prototype of your system in Python? I probably didn't understand the entity relationships in your explanation, so excuse my reaction. I hope you will see the problem from my POV. Good luck. V -- I do not respond to top-posted replies, please don't ask |
| drew@furrfu.invalid (Drew Lawson): Mar 11 07:13PM In article <mdpq3k$brm$1@speranza.aioe.org> >In CGI, the POST-data comes from the standard input, >so simply use fgets/fread/etc on standard input. The content of the post is on the standard input, but the size in in the HTTP variable CONTENT_LENGTH. That is retrieved as an environment variable. The CGI spec does not require end-of-file at the end of data. I know from experience that Apache 1.3 did not do that. I've never tested whether Apache 2 does. -- Drew Lawson | Pass the tea and sympathy | for he good old days are dead | Let's raise a toast to those | who best survived the life they led |
| You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.lang.c+++unsubscribe@googlegroups.com. |
No comments:
Post a Comment