201

Re: Application field With ++

Hello, Mystic Artifact, you wrote: MA> it is simple for completeness - C# has fair pointers. Unique here distinction - their culture not-use. In a C ++ we deal with them is more often, therefore is finite potential errors more. But the basic differences are not present. It not completeness, and errors. Pointers in Sharpe are completely not high-grade. They are strongly restricted. They can be used only in a special mode and only the marked area. Well, and the main thing that they are absolutely not necessary for writing of programs. They are used only as an optimization appliance. In 7th version created the safe pointer. They too it is strongly restricted. But reduce need for pointers even more. So to compare it it is not necessary.

202

Re: Application field With ++

203

Re: Application field With ++

Hello, prezident.mira, you wrote: VD>> 1. Rust. PM> It is true, I like teacher's tone. PM> but while rather early. You will wait all life and will say "while rather early". I already on the same D watched it. VD>> 2. D. PM> it is ridiculous. Ridiculously it is mentor tone and categorical announcements. Meanwhile D possesses everything that is in With ++ and much that in it is not present. And the main thing that on it where easier to write the code. During the last years With ++ rupture was a little tightened also is not become so obvious. But here already more than 10 years such as you use poor enough With ++ sentencing "while rather early". And now here there were also those who speaks "ridiculously". And without uniform argument. VD>> 3. Ocamle. PM> Tredov sinks, throw to it a life buoy more likely! It 25 sinks years. All in any way sinks. And different shouters not written in other languages code lines hang labels "sinks", "ridiculously", "rather early"... VD>> 4. Creation DSL generating the same / ++. PM> It is possible and from Scheme to generate a C. Only productivity at this code will be not as at manually written C, and as at a python. You at first itself try, and then . To a word I third of life am engaged in it. A difference between manually written code and generated I do not watch. It always can be made similar. Well, also the ideal code is not necessary everywhere. And to the Python can lose on speed unless... ... Well, Cut, i.e. same in a board  language.

204

Re: Application field With ++

Hello, prezident.mira, you wrote: S>> When thin manual tuning it becomes on a C ++ the same as and on a C is necessary. PM> Doyo For the sake of justice, in a standard C is restrict, and in a C ++ - is not present. (However, all necessary compilers understand __ restrict) About it was already told in the given subject (in other branch) the Author: so5team Date: 27.06 19:12.

205

Re: Application field With ++

Hello, VladD2, you wrote: N>> Yes, costs. In the same OpenCV which involuntarily became an example of subject library on With ++, too is expression templates for matrixes. I can write readable expression for any mathematician of type: Y = A * X + B. And the right part will be fulfilled by ONE (!!!) specialized function. Here it also is application With ++ which refines understanding of the code, does it easier without damage to speed. VD> it is possible not only on With ++. Nobody argues. But it also testifies that With ++ is modern enough and gives FREE in respect of productivity of abstraction of high level. About it also we speak.

206

Re: Application field With ++

Hello, Nuzhny, you wrote: N> Nobody argues. But it also testifies that With ++ is modern enough and gives FREE in respect of productivity of abstraction of high level. About it also we speak. 1. They completely not free. If it is necessary to squeeze out the last copecks in style With or on the assembler always it is possible to make faster. Simply the result such is not necessary to optimization of candles. But the same argument is true in 90 % of applications and to  to languages, whose compilers generate less optimized code. 2. It does not testify to the present. MT hardly is less than years than to programming languages generally. Lisp about 60 years ago was the first. 3. It is necessary to pay not only speed of performance of the code, but also simplicity, speed and convenience of development. In this plan With ++ does not merge unless bare With and to Pascal  flood.

207

Re: Application field With ++

Hello, VladD2, you wrote: VD> In the same toys norm began to write on With ++ bowels of engines, and games to describe on  scripts (in which about productivity even ridiculously to speak). Already one million years ago scripts on lua is actually luajit, which fast. I remember, in the first Stalker (many years ago) it was. So it is not necessary here about productivity of scripts. It normal.

208

Re: Application field With ++

Hello, VladD2, you wrote: VD> 1. They completely not free. If it is necessary to squeeze out the last copecks in style With or on the assembler always it is possible to make faster. Simply the result such is not necessary to optimization of candles. But the same argument is true in 90 % of applications and to  to languages, whose compilers generate less optimized code. Interesting, and what overhead projector in respect of productivity is imported by the same templates? It seems to me that you another answer something. VD> 2. It does not testify to the present. MT hardly is less than years than to programming languages generally. Lisp about 60 years ago was the first. And? The word modern is not to a synonym to a word innovative. All seems to me that you not with me argue, and with someone another. VD> 3. It is necessary to pay not only speed of performance of the code, but also simplicity, speed and convenience of development. In this plan With ++ does not merge unless bare With and to Pascal  flood. Aha. I write, means, operation with matrixes. It is necessary to me to count Y = A * X + B. And I on With ++ and write down! You represent? Also what as a result? In summary function which simultaneously multiplies is caused in me and adds matrixes. And, the type of elements of a matrix can be any. If the library is collected with usage Eigen its implementation will be caused. If the library has been collected with usage IPP this implementation will be caused. If has been selected OpenCL expression will be counted on the videocard. And me, the application programmer, it is possible to write simply Y = A * X + B. All! Where here a problem in simplicity, speed or convenience of development? Where it: "Au!" Now I can expect from you a similar example in other language?

209

Re: Application field With ++

210

Re: Application field With ++

Hello, prezident.mira, you wrote: PM> Without imbibed HKT - it is not necessary.  is Not the Kosher Companion? PM> it is not necessary. You here with  came? Very much is evident.

211

Re: Application field With ++

Hello, Nuzhny, you wrote: N> Already one million years ago scripts on lua is actually luajit, which fast. I remember, in the first Stalker (many years ago) it was. So it is not necessary here about productivity of scripts. It normal. The code on JS executed by means of JS-jit, works much faster, than the code lua on lua-jit. At Lua there is exactly one advantage - its interpreter (instead of the jit-compiler) is the champion on small consumption of storage. For this reason lua love for example in a telephony where it is necessary  to execute scripts (and normal jit there did not deliver because itself telephony branch scatter-brained) in one hundred.

212

Re: Application field With ++

213

Re: Application field With ++

Hello, Glory, you wrote: the Code on JS executed by means of JS-jit, works much faster, than the code lua on lua-jit. It is good that the big corporations could overtake one developer working for . But! That fact that JS-jit faster, than lua-jit does not say that lua-jit the slow. At Lua there is exactly one advantage - its interpreter (instead of the jit-compiler) is the champion on small consumption of storage. For this reason lua love for example in a telephony where it is necessary  to execute scripts (and normal jit there did not deliver because itself telephony branch scatter-brained) in one hundred. It agree. But in games I saw in delivery luajit is more often. Probably, for them storage is not strongly critical.

214

Re: Application field With ++

215

Re: Application field With ++

Hello, VladD2, you wrote: VD> 1. They completely not free. If it is necessary to squeeze out the last copecks in style With or on the assembler always it is possible to make faster. About With simple not truth. The code on With ++ always not more slowly than on S.Esli it is finite at the programmer of a hand not curves. Simply because templates are optimized much better, than the procedural code. A classical example qsort vs std:: sort. On the assembler in the theory it is possible, but in practice very few people can overtake the modern compilers.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

216

Re: Application field With ++

Hello, Nuzhny, you wrote: N> Now I can expect from you a similar example in other language? On  and any language with similar macroes such without problems becomes. And macroes much more abruptly templates.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

217

Re: Application field With ++

Hello, WolfHound, you wrote: N>> Now I can expect from you a similar example in other language? WH> on  and any language with similar macroes such without problems becomes. WH> and macroes much more abruptly templates. Well, well. But examples will not be, yes?

218

Re: Application field With ++

Hello, Nuzhny, you wrote: N> it is good, good. But examples will not be, yes? About matrixes will not be. For that they were them someone should write. But for that understanding that at it probably look on http://www.nemerleweb.com/tutorial This piece takes the code on  and transforms it into the reactive code on JS. Or here on it http://omega.sp.susu.ru/books/conferenc … ll/117.pdf This piece takes the code on  and transforms it into the code for the videocard. In both cases there is considerably more a complex analysis and a code rewriting.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

219

Re: Application field With ++

Hello, Mystic Artifact, you wrote: WH>> Where access still a question is better. WH>> http://rsdn.org/forum/nemerle/4220330.flat the author: Denom Date: 04.04.11 MA> you are not absolutely clear on what gave the reference (that was specific meant). . On the code which soils With ++ on productivity. The remaining I will not answer, for you answered the imaginations which do not concern that on what I gave the reference.... <<RSDN@Home 1.0.0 alpha 5 rev. 0>>

220

Re: Application field With ++

Hello, WolfHound, you wrote: WH> Or here on it http://omega.sp.susu.ru/books/conferenc … ll/117.pdf WH> This piece takes the code on  and transforms it into the code for the videocard. About web I will tell nothing, because I do not understand. And about NUDA it is possible. As practice shows, on the same OpenCL it is necessary to write the individual code almost on each architecture. In general: AMD VLIW - one code, GCN - another, for NVidia - the third, under Intel - the fourth. It is the minimum, actually it is necessary to do more gradation to receive a productivity maximum. Someone declares that supports standard OpenCL 1.2, but in practice it appears that some functions from the standard do not work, return , but anything thus do not do. And it is normal in this area. OpenCL on mobile platforms in more pitiable state. And still everyone  has the extensions much of them are registered in the standard, but not all. So to me it is not especially trusted in possibility qualitative  OpenCL the code. All the same article resulted by you old enough, the modern realities is more difficult. Whether it is necessary to say that in the same OpenCV over OpenCL and CUDA the code regular employees from Intel,  and Nvidia worked. For all is difficult, at  it turns out to write much more productive code because of presence of the internal detailed documentation. And some things at all . Therefore that on with ++ that on Nemerle all the same it is necessary to write not generalized, and the dedicated code. Here no scoring will exist. And it is necessary at level above, level of the application programmer which uses language, library and can write on close to itself language. In my context is language closer to mathematicians. It is possible to be guided by Matlab who is de facto the standard for them. Here the client code on OpenCV on With ++ can be not much more difficult  and . I somehow took small function non maximum suppression on the Python (from the person which sells the lectures about computer sight on the Python) and rewrote it on With ++. It turned out more shortly and more effectively even in respect of algorithms, not to mention total productivity. With ++ + OpenCV it appeared more expressively, than the Python + numpy + OpenCV (official  to the Python). Here it also is simplicity and efficiency.

221

Re: Application field With ++