C++ :: Struct Alignment Across Compilers And Operating Systems
Oct 5, 2013
I am working on extending an application that features a (rather old) plugin system. The plugin system is designed to work on x86 only, but support any and all compilers on windows, linux and also OSX.
My goals are now to
1) extend the plugin framework in a backwards compatible way
2) port it elegantly to x64
So far so good.
The plugin subsystem essentially exports a bunch of SDK function like this:
Code:
extern "C" CDECL int somefunc(int a, int b);
Which can be used by the plugin, and expect the plugin to export some function in return. When a struct is supposed to be send to the plugin, the syntax of the system goes like this:
Code:
extern "C" CDECL int RequestStructField(int field, int structhandle);
So the struct isnt handed over via a pointer, but every field has to be requested individually! From what I can tell from the comments, the developer was afraid there might be problems exchanging structs between different compilers, alignment and all.
I would like to tear down this limitation, and pass pointers to structs to plugin directly. Or better: allow the plugin to give me a pointer to an empty struct which i subsequently fill.
However, i have no control over the compiler or language used to write the plugins. Quiet some plugins are written in Delphi, some even in MASM ... both would have to support this way of passing structs.
the question is: does that work?
I have looked at other plugin systems, but they either use entirely different ways (XPCOM) or are limited to a single platform or compiler.
Some of them have a note stating that the plugin must be compiled with "byte-aligned" structs.
x86 and x64 plugins are two different stories, x64 version of the app does NOT have load x86 plugins. I dont expect there to be too man problems (right?). But we'll see, i guess.
I work on a system that communicates with other systems via messages. Those messages are defined in a spec and every word must be exactly as defined. To accomplish this Ada allows me to define the fields of my record to specific words in memory.
Code: word = 4; type Msg_Type1 is record x: Some_Type; y: Another_Type; z: Another_Type; end record;
I'm having a problem with the strcmp return value. My compiler returns the correct value based on my research, but I tried to compile the code below on those online compilers and the result was 1.
My compiler gave me 10, which I consider to be the correct result according to strcmp - C++ Reference. So strcmp is not very portable?
Basically, I'm working on a code that has its own allocator class. In the backend, the class uses malloc() to create huge blocks of memory aligned to sizeof(void*).
Apparently, in certain special circumstances, sizeof(void*) will be 4 when the data actually needs to be aligned to 8 bytes. If the memory isn't aligned to 8 bytes, there will be heap corruption and a bus error will occur. One such case is on HP-UX; we forced the alignment to 8 bytes on that platform. According to the commit log, HP-UX has a minimum struct size of 8 bytes and that's why it has to be aligned to 8 bytes (but I can't confirm if this is actually true).
We're running into these same problems when compiling on MIPS, MIPSel, and SPARC (all 32-bit builds).
(Side thought just now as I wrote that: seems like RISC processors don't allow unaligned access while x86 processors do. It looks like these build failures on MIPS and SPARC were while trying to build 32-bit on 64-bit hardware, could this be the problem?)
Anyway, the ultimate question and goal here is to find out how to detect this magical minimum alignment size, if not sizeof(void*).
I want the function to display all accounts deposit list to work properly, what i mean is that the table coming has its contents here and there . I want to align them properly. program in comment.
how to ask the user to input operating symbols like "+","-","*" and then use it to operate on the numbers. The user should be asked to input the symbols not like this
I developed one windows application in c#.net , this application is not working when i installed the setup in another system with another sql database server-r2. How can solve this. here i had attached my code for reference.
I'm currently writing a cross-platform on screen keyboard application that highlights keys on the screen as you type them. My approach was to use an image of a keyboard and layer semi-transparent sprites over each key on key events, and as a consequence I needed to map out precise positions of each of these sprites relative to the background image.
The issue became how to load those positions into my application during runtime? My solution was to use excel to make a .csv of all the x-y coordinates, but instead of just deploying the .csv with my application I decided to write a quick app that parsed each line of the csv for coordinate pairs and write those floating point numbers to a binary file instead. That way the end user can't easily edit the coordinates of the sprites.
Now on to my question. The way I load the data into my application is by this method:
#include <fstream> #include <vector> /* other includes */ /* ... runtime code ... */ std::vector<float> pixmapPosData;
[Code] ....
This works fine on my system, but what if sizeof(float) has a different length their system? Then this function could throw a read access violation towards the end of the file (since the .bin file is exactly 508*sizeof(float) bytes long on my computer), and not to mention the coordinates would be meaningless. Is there any way to avoid this? Or maybe I just just suck it up and deploy a .csv file?
I'm taking an operating systems class and I'm trying to write a script that would execute a program in Simple OS.I"m not sure if this is the right place for this because although the entire OS is written in C, "asm" are assembly language code. Anyway when I run the OS in virutal box and try to run a program, I get a "Fatal Execption" in a loop until the entire OS crashes. So then I try to debug my code BUT when I add printf statements and try to run the code again I don't get to the "fatal execption" and the entire OS just crashes. 1) why can't I add print lines? 2) what's going on with the fatal execption?
//////////////////////////////////////////////////////// // Loads program and executes it in user mode // #include "kernel_only.h"
I'm trying to find out, how to do printf so that it prints out text in the center of the window. I know how left and right justifyment is... but how to get to the center? C...
I was attempting something weired with address to move data around when I discovered that the size of the array is not what I expected. I am passing this structure as &Users to a function that declares it as a void *, then I can deal with chunks of data (memmove) and not have to worry about index or things like that. However...sizeof is returning something I do not understand.
Im having trouble creating a struct within a struct node. the program suppose to hold students firstname, lastname, and gpa in a node therefore creating my linked list. Line 26 keeps saying that cannot convert parameter 2 from 'studentType to std::string
error: request for member 'character' in '* ptr', which is of non-class type 'datastructure*' error: request for member 'character' in '* ptr', which is of non-class type 'datastructure*'
These errors are related to " *ptr->character='a'; printf("Ptr: %c",*ptr->character); "
I want to access "character" data inside the structure "trial" by a pointer to pointer "ptr" inside function "function",but I couldn't find a way to do this.
I just read and have known for a while that classes are private (members and inheritance) by default and structs are public. But my question then comes what if.. a struct inheriting from a class or a class inheriting from a struct?
struct example { int x; int y; }; int main() { example variable; variable.x = 1; variable.y = 2; map<example, int> something; something[variable] = 3; return 0; }
And I get a very long error message from the compiler. I tried using an enum as the key as well and that worked just fine, but I guess that's just the same as having an integer as the key. But anyway, what's the problem with this? Is there any way to make it work?
So i create a simple structure,that has a firstname, lastname, age, dateofbirth.I would like it if i type what I am searching for print out there is one or not.But i have some trouble with equals(in array).. i type in the same "keresettnev" as "tanulok[i].Firstname" but it thinks not the same. (it works with age..)
#include <stdio.h> struct database { int id_number; int age; float salary;
[Code] ....
When I compile, I get an error: test.c|18|error: incompatible type for argument 1 of 'printf'| note: expected 'const char *' but argument is of type 'float'|
I thought employee.salary is a float but the compiler expected 'const char'. How do I make this work?
I am working on a program which uses external hardware to acquire data. we have the option to use hardware from two different companies, each with it's own driver. However both do the same job. My program is meant to read data packets which are structured as:
I am programming my software so as t allow the customer to use the hardware of their choice. Simply select the card they are using and our program should be able to take care of the rest of stuff. I am using #ifdef directives to include the header for the corresponding hardware dll. Now I want to define my own data struct of the type:
I've been working on a path-tracer for some time, and all along I've used structs instead of classes for vectors and matrices. Today I changed all of them to classes instead, changing none of the actual function bodies themselves, and the results were miserable to say the least.
Here's a render form before the change: [URL] ....
And here's the same render after: [URL] ....
Why this is happening, considering that none of the actual function-bodies have been changed, except for what little is needed to make the change from class to struct.
So this is the last part of a program I've been working on for four weeks now. This question may be a tough one considering the amount of files included in the program.
The program is to read in a file of requests between two cities, read in a file of flights and cities that occur between the cities. It then checks to see if there is a path between the flights and output an itinerary. I have the correct itinerary outputting, but when attempting to output the associated flight number and price according to the city, I am getting odd data. How can I output the correct flight number and price associated with each flight on the itinerary.
I'll post out the output I am currently getting and the section where I am outputting the data. I'm sure I'll need to post more files so the program can be understood.
Don't want the code done for me, just a point in the right direction! I don't want to let this program defeat me!
Output: Code: Request is to fly from Atlanta to San-Diego.The flight itinerary is: Flight # From To Cost 10 Atlanta Chicago $134529069 10 Chicago Miami $134529069 10 Miami Dallas $134529069 10 Dallas San-Francisco $134529069 10 San-Francisco San-Diego $134529069 This function finds a path between cities. Code: bool flightMap::IsPath(string originCity, string destinationCity){ StackClass aStack, bStack; flightStruct flightRec; string topCity, nextCity; bool success; int index = 0;
I am working on an assignment identical to another post from a couple years ago, for reference here is the thread:
array of pointers to structures sorting addresses by zip code
They way it is written on that thread is almost identical to the way the teacher implied to have it done (only wrote part of the input block). But I am having an error:
When it gets to the output section it outputs then next name along with the zip code... I tried strncpy and strxfrm but both cause more problems than they did work.
The last part of the project is to have the output put out in order of least zip code to most zip code (00000<99999), so this is causing me a real problem and I do not see what exactly is making this happen.
Here is my code (we dont HAVE to use gets but professor suggested using it for this assignment, next lab is to rewrite this using files and fgets rather than I/O redirection):
I have not started the sorting code because I cannot get past this, but once I have proper zip codes I am sure I can make a sort function no problem.
I am using xcode with some breaks to read variables as various points and do not notice anything wrong until it makes it to the output functions, although this page briefly pops up between input and output functions when the breaks are up:
