.net - C# fundamentally not portable? -

-

i've been using c# while, , have started working on adding parallelism side project of mine. so, according microsoft, reads , writes ints , floats atomic

i'm sure these atomicity requirements workout fine on x86 architectures. however, on architectures such arm (which may not have hardware floating point support), seems these guarantees hard.

the problem made more significant fact 'int' 32-bits. there many embedded devices can't atomically perform 32-bit write.

it seems fundamental mistake in c#. guaranteeing atomicity of these data types can't done portably.

how these atomicity guarantees intended implemented on architectures there no fpus or 32-bit writes?

there 2 issues regard "portability":

  1. can practical implementation of language produced various platforms
  2. will program written in language expected run correctly on various platforms without modification

the stronger guarantees made language, harder port various platforms (some guarantees may make impossible or impractical implement language on platforms) more programs written in language work without modification on platform support exists.

for example, lot of networking code relies upon fact (on platforms) unsigned char 8 bits, , 32-bit integer represented 4 unsigned chars in ascending or descending sequence. i've used platform char 16 bits, sizeof(int)==1, , sizeof(long)==2. compiler author have made compiler use bottom 8 bits of each address, or have added lot of code writing 'char' pointer shift address right 1 bit (saving lsb), read address, update high or low half based upon saved address lsb, , writing back. either of approaches have allowed networking code run without modification, have impeded compiler's usefulness other purposes.

some of guarantees in clr mean impractical implement in platform atomic operation size smaller 32 bits. what? if microcontroller needs more few dozen kbytes of code space , ram, cost differential between 8-bit , 32-bit pretty small. since nobody's going running variation of clr on part 32k of code space , 4k of ram, cares whether such chip satisfy guarantees.

btw, think useful have different levels of features defined in c spec; lot of processors, example, have 8-bit chars can assembled longer words using unions, , there lot of practical code exploits this. define standards compilers work such things. see more standards @ low end of system, making language enhancements available 8-bit processors. example, useful define overloads function can take run-time-computed 16-bit integer, 8-bit variable, or inline-expanded version constant. often-used functions, there can big difference in efficiency among those.


Comments

Post a Comment

Popular posts from this blog

C#: Application without a window or taskbar item (background app) that can still use Console.WriteLine() -

-
i'm trying create application adheres following: no taskbar no console or form window can utilize console.writeline() . (i.e. if executes app command prompt write console.) the problem if create windows form (or wpf) application can have there no taskbar, console or window show up, console.writeline() nothing. if create console app, writes console, can't figure out how hide (and if did hide it, write command prompt window?)... how do this? just create standard console application. callers responsibility hide window caused program (from how run c# console application console hidden ): system.diagnostics.processstartinfo start = new system.diagnostics.processstartinfo(); start.filename = dir + @"\myprocesstostart.exe"; start.windowstyle = system.diagnostics.processwindowstyle.hidden;
Read more

c++ - Convert big endian to little endian when reading from a binary file -

-
this question has answer here: how convert between big-endian , little-endian values in c++? 28 answers i've been looking around how convert big-endian little-endians. didn't find solve problem. seem there's many way can conversion. anyway following code works ok in big-endian system. how should write conversion function work on little-endian system well? this homework, since systems @ school running big-endian system. it's got curious , wanted make work on home computer also #include <iostream> #include <fstream> using namespace std; int main() { ifstream file; file.open("file.bin", ios::in | ios::binary); if(!file) cerr << "not able read" << endl; else { cout << "opened" << endl; int i_var; double d_var; while(!file.eof()) {...
Read more

openssl - Load PKCS#8 binary key into Ruby -

-
i'm trying load particular private key encoded in binary der format ( pkcs#8 ) ruby. however, openssl::pkey won't recognize it. can make work doing console work , transforming pem so: openssl pkcs8 -inform der -in file.key -passin pass:xxxxxxxx >private_key.pem after this, key can correctly read. however, since whole process done in memory instead of writing , reading files. so question is: possible load private keys binary encoded der format ruby/openssl? thank time, fernando yes, can indirectly load pkcs#8 der-encoded private keys using ruby openssl. openssl::pkey::rsa.new handle pem-formatted pkcs#8, easy read binary der , convert pem-formatted string , load string. for example, these der-encoded private keys: $ openssl genrsa | openssl pkcs8 -topk8 -outform der \ -nocrypt -out pkcs8.key $ openssl genrsa | openssl pkcs8 -topk8 -outform der \ -v2 des3 -passout pass:secret -out pkcs8_des3.key you can this: require 'open...
Read more