Ensure the syntax is what we guess

Question

Raphael Cautain 2012-8-17

2
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/46227-ensure-the-syntax-is-what-we-guess

It would be nice to publish (as in Old School Languages) a syntax summary of Matlab, reflecting the exact behaviour of the interpreter.

It could help to prevent surprise from unexpected results, such as when one evaluates the expression : 5 + 1*0:10

A single transcription of the formal grammar contained in the interpreter's code would be sufficient.

Thanks.

2 个评论
显示无隐藏无

Walter Roberson 2012-8-17

Yes, a BNF would be good.

Walter Roberson 2012-8-17

I would not call it a bug. It just reflects that ":" is a low priority operator, which is consistent. If you want to raise the priority of the ":" then use brackets.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Jan 2012-8-17

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/46227-ensure-the-syntax-is-what-we-guess#answer_56536

You do not have to _guess_ the details of the syntax. It is explained exhaustively in the documentation already. Your example |5 + 1*0:10| is clear after reading:

doc precedence

2 个评论
显示无隐藏无

Raphael Cautain 2012-8-18

That's right, thanks. This precedence rule is very useful when we write subscripts, like in "subimg = img(x-r:x+r, y-r:y+r);" We just lost the equality of sizes between 0:10 and 1+0:10, which is really less useful and requires then to write 1+(0:10). No problem, it's just an habit.

Jan 2012-8-18

Note, that X((1+0):(1+10)) will be processed faster than X(1 + (0:10)): In the 1st case the limits are checked for the marginal elements only, while in the 2nd case Matlab wastes time with boundary checks. Of course this is more important for larger index vectors.

请先登录，再进行评论。