1

Topic: Bones - meat

I welcome. I do not have experience of joint programming and experience under the direction of the developer, therefore, possibly, the idea is silly. But very much it would be desirable to ask. When I worked as the programmer, I very much would like two things: 1. That there was more knowing developer who would give me titles of classes which I could . 2. That there was less knowing developer to whom I could give titles of procedures/functions which it could . In other words: 1. The architect  the future classes and their interaction 2. , in these classes,  properties/methods 3.  filled properties/classes. So do?

2

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> So do? No because it's totally ludicrous from discharge "the master  nuts - the trainee twists". Was ineffectively for everything, in any way does not promote training and in any way does not reduce risk of errors. Micromanagement happens is useful to input absolutely green  in a command, its main objective - to disassemble all unobvious nuances of operation in practice. In most cases there affairs on pair-triple days. Further -

3

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> In other words: R3> 1. The architect  the future classes and their interaction R3> 2. , in these classes,  properties/methods R3> 3.  filled properties/classes. R3> so do? No so normally do not do. Development of the interface of a class and its implementation almost always become simultaneously. But, at the same time, there is a restricted circle of tasks where they need to be developed separately (for example, implementation of the new component implementing  API, in that case if API idle time and its implementation is laid down in one class. It is possible to give incidentally still to another, for example, interface implementation (if many implementations are necessary, and time has not enough), but here there is a risk  which needs to be considered.

4

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> In other words: R3> 1. The architect  the future classes and their interaction R3> 2. , in these classes,  properties/methods R3> 3.  filled properties/classes. R3> so do? Absolutely so do not do. Something similar can be when it is a question of changes in API.

5

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> In other words: R3> 1. The architect  the future classes and their interaction R3> 2. , in these classes,  properties/methods R3> 3.  filled properties/classes. R3> so do? Do everywhere, except for absolutely small  and huge  monsters where fight and  games. Normally  or the skilled developer covers with structure of classes/interfaces and the signature of key functions, writes a unit tests and gives  . It fine saves time of expensively developer at the expense of cheap . Naturally, there are tasks where qualifications most  insufficiently,  or refactorings, such operations it does everyones.

6

Re: Bones - meat

Hello, pestis, you wrote: R3>> So do? P> do everywhere... You in minority.

7

Re: Bones - meat

Hello, Sinix, you wrote: S> Is not present, because it's totally ludicrous from discharge "the master  nuts - the trainee twists". Was ineffectively for everything, in any way does not promote training and in any way does not reduce risk of errors. Dispute. And I think, as to training and errors, so completely on the contrary: while I , what classes are necessary to me, I will lose so much time and I will do so much errors that oh-oh th. Otherwise - I will strike I see as correctly. "The master  nuts - the trainee twists" - for example, wheel nuts are recommended to be fastened in certain sequence; the first time without the master in any way; when still the trainee learns about it.

8

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> Dispute. And  R3> And I think, as to training and errors, so completely on the contrary: while I , what classes are necessary to me, I will lose so much time and I will do so much errors that oh-oh th. Otherwise - I will strike I see as correctly. Not, it is clear that training is necessary for elements, but 1. It happens in the beginning, speech about  (at  about it) to this moment does not go. 2. Training never is under construction by a principle "I here the code wrote - complete". At first we acquaint with agreements in a command, then simple real , in which corrections - one line to change, then hardly more difficult, but too typical, and  if the person does not consult - what for him to torture? If consults - problems with passage to independent implementation of simple features and, at last, to independent operation under the user scenarios does not arise any more. Well and certainly, at each stage remaining members of team work not by a principle "do as I told", and help to learn to understand with problems independently. Otherwise in any way - sense to raise eternal ?

9

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> In other words: R3> 1. The architect  the future classes and their interaction R3> 2. , in these classes,  properties/methods R3> 3.  filled properties/classes. R3> so do? Earlier such model saw regularly. Recently practically did not meet. It is peculiar for projects which go on waterfall in the companies where there are selected architects and application designers.

10

Re: Bones - meat

Hello, Sinix, you wrote: S> 2. Training never is under construction by a principle "I here the code wrote - complete". This too categorical statement. Actually it is one of possible approaches to training. S> at first we acquaint with agreements in a command, then simple real , in which corrections - one line to change, then hardly more difficult, but too typical, and  if the person does not consult - what for him to torture? And whence typical  undertake that? It in the core various  when skeleton of a feature is ready, were operations on a finishing-polish-finishirovaniju.

11

Re: Bones - meat

Hello, Real 3L0, you wrote: R3> I do not have experience of joint programming and experience under the direction of the developer, therefore, possibly, the idea is silly. But very much it would be desirable to ask. R3> when I worked as the programmer, I very much would like two things: R3> 1. That there was more knowing developer who would give me titles of classes which I could . R3> 2. That there was less knowing developer to whom I could give titles of procedures/functions which it could . Looking what purpose. If control I pound is not enough. And if training it is a good variant - one does sketches, another implements them. But if   it is necessary to go a similar way, only to begin not with class titles, and from the causing code. If speech about control it makes sense to change a little the approach - the leader implements the main user-case, minor  the conducted. Here a difference that not so much titles, how many interaction will be ready. This part, interaction and behavior, is most difficult for perception less  the developer. This part, instead of classes-functions-modules is actually projected. Actually it is necessary to divide not on more or less knowing, and on conducting-conducted. In different parts of the project they even can change over. The principle remains without changes.