1

Topic: The organization of source codes

2

Re: The organization of source codes

3

Re: The organization of source codes

Hello, GhostCoders, you wrote: GC> and to use this library on what-nibdu old CentOS 5.x or even CentOS 4.x. Requirements libmuslc.so to OS are minimum - kernel Linux 2.6 or above, GC> there are no dependences on it glibc. And generally, such.so it is possible to use practically on any modern Linux (with a kernel 2.6 or above). That is for support of "all distribution kits Linux" it is enough to collect the closed library on Alpine Linux c Beautiful Capi.

4

Re: The organization of source codes

Hello, GhostCoders, you wrote: GC> Requirements libmuslc.so to OS are minimum - kernel Linux 2.6 or above, GC> there are no dependences on it glibc. Thanks, a worth-while experiment. Unfortunately, static  still it can be not so applicable because of Copyleft, in particular, GPL. If  are given on sale, we receive legal problems. And here it looked? http://refspecs.linuxfoundation.org/lsb.shtml if briefly this requirement, what libraries is mandatory should be in each distribution kit for binary compatibility. In my case while it works, but experience for creation transferred  for Linux at me absolutely small so I can not know about any other reefs. Just in case, all I compile in CentOS, there C means ++ suffice 11 me. Tried on Ubuntu on most , it is reverse  work,  writes in  the version  as compatible. CUDA it is possible  statically, for OpenCL it was necessary to write the .so so and to search on file system.

5

Re: The organization of source codes

Hello, rean, you wrote: R> GCC and MSVC from IDE throw all garbage of compilation a little differently. (About XCode anything I can not tell, since I do not use.) GCC litters directly near to a source code, R> MSVC creates subdirectories and all throws there. In IDE and systems of the assembly it is adjusted.

6

Re: The organization of source codes

Hello, GhostCoders, you wrote: GC> In Linux the situation is a bit worse. GLIBC (with Runtime Library) - the analog of Microsoft C Runtime Redistributables (VC_Redist.exe) - there is beaten by nails in OS. GC> to use static glibc - it is impossible (cheto it is not finished), especially it is impossible to use static glibc in dynamic divided (.so - shared object) library. GC> I found a way out in  Alpine Linux - in this OS is used the implementation of C Runtime library - muslc. GC> But  static libc to shared object - again  it is impossible. Strange. I so deeply did not dig, and  glibc statically both to programs, and to.so. Problems was not, but it was transferred between different

7

Re: The organization of source codes

Hello, rean, you wrote: R> Further,  with development of template programming more and more source codes now I watch only in . , but the principal reason of library without compilation units, it that libraries in the form of heading files are easier for using, and in all the zoo of systems of the assembly is guilty. R> further on file system. R> GCC and MSVC from IDE throw all garbage of compilation a little differently. (About XCode anything I can not tell, since I do not use.) GCC litters directly near to a source code, the Nonsense any, gcc is the console utility, it has a flag "-o" and she writes down result of operation where it is necessary. R> MSVC creates subdirectories and all throws there. It caused certain inconveniences at testing and prototyping as the current directory thus is not near to exe. R> Badly and there and there, I made differently, thanking build to system. R> my choice fell on ninja build. There were alternatives: Makefile, CMake and others. It is very strange to compare the incomparable. cmake it is the generator of design files, including he is able to generate ninja files. R> it is finite, for each compiler it was necessary to write pens all from zero, but now I am assured that all is made optimally, after all I have a complete control over files and compilation keys. R> and yes, I support GCC, MSVC, CUDA Nvcc, Spir and a heap , including and . And what for? cmake, bazel all are able too most, and it is possible to write type if gcc flags such if msvc flags such too there will be one line on flags for the compiler, too will be generated ninja files, what for to pay more if all is already ready? R> development as a whole. R> at me on different disks in different sections on different computers the heap of different operating systems is delivered. R> I Am overloaded and I work there. It is necessary to test in other system - I am overloaded and again I work there. R> (By the way, advise convenient and simple editors for Linux GUI and MacOS. Conveniences Windows do not suffice me on Linux and a poppy.) editors or IDE, the part of your problems which you can solve disappear, for example with the organization of files and other if we tell will use IDE CLion on all OS?

8

Re: The organization of source codes

Z> the Nonsense any, gcc is the console utility, it has a flag "-o" and she writes down result Z> operations where it is necessary. I wrote not about physical possibility or impossibility, and that associates with that or other compiler. Also it is a lot of the projects written with GCC, and gathers - all in a heap rushes. Can be to me so got, but the most part of all indirect that I collected with GCC, and gathered. R>> and yes, I support GCC, MSVC, CUDA Nvcc, Spir and a heap , including and . Z> And what for? cmake, bazel all are able too most, and it is possible to write type My tool to me approaches more. Z> too there will be one line on flags for the compiler, too will be generated Z> ninja files, what for to pay more if all is already ready? What for all this superfluous though it is possible to use directly and with direct control over literally each character? There there are advantages, what for to use interlayers, but my tool approaches me more. Just payment does not happen any more, there is no all superfluous and intermediate. The price of such flexibility - it is necessary to track all independently. In what and convenience of my choice. I supervise all with minimum efforts and with a minimum dial-up the tool. Ninja build is one small file in all systems. And it is written not on scripts, does not demand setting  languages, something else does not demand in dependences. Does not climb in registers, does not search for compilers and does not refuse to work with that that is not installed. Is able to see dependences and it is correct to collect /C ++ projects if changed any . R>> (By the way, advise convenient and simple editors for Linux GUI and MacOS. Conveniences Windows do not suffice me on Linux and a poppy.) Z> Editors or IDE, the part of your problems which you can solve disappear, Z> for example with the organization of files and other if we tell will use IDE CLion on Z> all OS? At present to me it is inconvenient to time to edit in a quickly way files in these systems. It would be desirable something type Notepad ++. Same superficial, and with syntax highlighting. If to launch something become serious, I use QtCreator.

9

Re: The organization of source codes

Hello, Marty, you wrote: M> it is strange. I so deeply did not dig, and  glibc statically both to programs, and to.so. Problems was not, but it was transferred between different  interesting. And it is possible for me to send any program and so? For learning, so to say.

10

Re: The organization of source codes

Hello, GhostCoders, you wrote: M>> it is strange. I so deeply did not dig, and  glibc statically both to programs, and to.so. Problems was not, but it was transferred between different  GC> interesting. And it is possible for me to send any program and so? For learning, so to say. The last time tried about five years ago, excuse, does not remain near at hand

11

Re: The organization of source codes

Hello, Marty, you wrote: M> the Last time tried about five years ago, excuse, there was no near at hand There like any a bug in glibc that supposedly at instructions usage static glibc, any libraries (something  libgcc_s.so) all the same  dynamic. I can is wrong. Still such singularity glibc it that any requirements with a kernel (more rigidly than the requirement of a kernel 2.6 are shown)