• 
• 

• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 

Spelling and punctuation

The spelling and punctuation is, certainly, care of the editor and the proof-reader. But in the specialized editions with literacy as you can see on many examples, at all so it is good, as in political and even popular scientific. The problem that the proof-reader sees many terms for the first time, instead of understands practically all. Therefore he or does not correct them, hoping on the editor, or - that is even worse - correct on own understanding. The editor, in turn, understands all, but hopes for the proof-reader. Comments in a code and to an illustration are especially subject to this misfortune - because him{it} perepadaet is attention. Everyone consider, what is it not by their part. The scientific editor counts, that everything, that in Russian is to literary, literary - that a code in conducting scientific, and illustrations is in general to the artist, the proof-reader circumspectly at all does not touch neither a code, nor pictures - from a sin far away. The result can be the extremely unfavourable. Therefore do not consider, that edition will correct all achepjatki and will put down all points. Recollect, that all the same you write on the native language, and consider it{him}, as native. By the way, "you" in Russian it is written from the big letter only at the personal reference{manipulation}.

Correctness of conclusions

Clauses{Articles} connected to programming are especially technical clauses{articles} if they, certainly, do not concern to section " Colleagues, smile ". Questions of belief in these clauses{articles} are inappropriate. If you do{make} any statement, it should be properly proved - by tests, gaugings or logic conclusions.

Tests

Many authors, carrying out in clauses{articles} of comparison of any products, approaches or algorithms, count the tests so trivial, that do not reflect at all on an opportunity of completely banal mistakes in these tests. Therefore they are not puzzled with check of the code. However practice shows, that probability of mistakes in, apparently, elementary cases, is rather high. Ability to make a mistake is inherent in each person, and exceptions here no. To avoid similar situations, it is necessary to check not simply serviceability of tests, but also to build them so that to have opportunities easily to check up correctness of given out results. For example, if you compare productivity of two means of development, say, compilers in integer operations, it is the best way to pick up the well-known algorithm giving on an output{exit} not less well-known result. So, the algorithm of calculation of number Pi obviously should give out number Pi, instead of something another.

At checks of result care too is necessary. Mistakes can contain not in algorithms, and in the accompanying code used by these algorithms. For example, if you test speed of sorting and use thus external objects or functions - comparators, the mistake can contain in one of methods of comparison. The check of result made with the help of the same comparator, most likely, will not reveal a problem. Therefore check of results is better for carrying out with the help of a separate code, whenever possible not using any external components. In the mentioned case of sorting optimum check will be the banal cycle checking, that each following element more previous, and calculating the control sum which can be verified with the control sum calculated for not sorted file.

Remember that not taken into account{not discounted} external conditions can influence the test. Not taken into account{not discounted} because the author at the moment of creation of tests did not guess their existence. Probably, you are surprised, but handicapes can appear such things, as the optimization, carried out by modern processors. For example, attempt to measure speed of a call of function of the interface by its{her} call in a cycle, most likely, will not give any reasonable results. The matter is that even if such call will not be eliminated by the clever optimizing compiler, his{its} speed will be affected all the same with the optimization which is carried out by the processor. Modern processors contain the mechanism of a prediction of the branchings, fine working at such tests, but powerless when the call of such function meets in a usual code. An output{exit} from this situation can be selection of the algorithm intensively using virtual calls.

In general, approach to the tests critically, try to check everything, even the most improbable, assumptions and, in the best way, create some variants of tests for one tested aspect.

Scale

If you have decided something to extol, or on the contrary, to consider essential lack, think, as far as it will affect an end result. Often there is so, that an author, having hooked for, as a whole, a unimportant thing, makes mountains out of a molehill, or representing her{it} as panacea, or, on the contrary, arranging above her a great court. Meanwhile, the fact can not cost vyedennogo eggs on a background of other merits and demerits of described technology, language or a platform.

Bicycle

" I have found how to apply here not erased trunks from

Polystructural fibre with vyrozhdennymi aminnymi communications{connections}

And incomplete oxygen groups.

But I do not know while how to use recycling

Reactor on subthermal neutrons. Misha, Mishok! How to be with a reactor? "

Having looked narrowly to the device, I have without effort learned{have without effort found out} a bicycle.

A. And B.Strugatskie, Monday begins on Saturday.

This nice{famous} device for moving to space was already mentioned above, but to repeat nevertheless costs{stands}. If you were occurred with ingenious idea which anybody has not guessed yet another, do not hurry up to create the treatise on this subject. Check up - probably, this idea has occurred not only to you. For example, state her{it} shortly in a forum corresponding on subjects - for example, in " Philosophy of programming " on RSDN. If the idea is not new, you learn{find out} about it literally for day - other - at us full wishing to shout "Bayan" at the slightest pretext.

That your idea appeared not only yours, does not mean yet, that your reflections have no any value. Often it happens so, that the person considering " from zero " something already known, finds own not ordinary sight at a problem. It can appear interesting to readers.

Be not afraid, that your idea will steal. If all of you are afraid of malicious competitors, patent idea. But in this case you all the same should prove its{her} novelty.:)

The conclusion

All resulted mistakes are so typical what to eradicate them will fail any clause{article}, to six hundred sixty four, mentioned in introduction. However if after this clause{article} to someone will come in a head before sending in edition to re-read clause{article} once again and something to correct in her, I shall consider the problem{task} vypolnenoj - in fact it even will a little reduce quantity{amount} of reader's memoirs on close relatives of the author, the editor and mine personally.