Re: question about linker

On 04/12/2024 22:01, Bart wrote:

On 04/12/2024 19:47, Janis Papanagnou wrote:

On 04.12.2024 13:08, David Brown wrote:

 

Sure.  And it is certainly /possible/ to know all the small details of C
without ever reading the standards.  But it's quite unlikely.

>
The question is (IMO) not so much to know "all" and even "all small"
details. Even in a language like "C" (that I'd consider to be fairly
incoherent if compared to other languages' design) you can get all
"important" language properties (including details) from textbooks.
>
If cases where that is different, the standards documents - which
have their very own special way of being written - would be even
less comprehensibly as they (inherently) already are. - That said
from a programmer's POV (not from the language implementors').
>
I look into language standards only if I want to confirm/falsify
an implementation; in this case I'm taking the role of a language
implementor (not a programmer). Personally I do that anyway only
rarely, for specific languages only, and just out of academical
interest.
>

[...]
Bart is an expert at thinking up things in C that confuse him.

>
Well, that's an own topic. - Where I was really astonished was the
statement of being confused about the braces/semicolons, which is
so fundamental (and primitive) but technically just a detail that
I'd thought it should be clear

  OK, if it's so simple, explain it to me.
 Apparently the first line here needs a semicolon after }, the second doesn't:
    int X[1] = {0};
   void Y() {}
 Similarly here:
    if (x) y;
   if (x) {}
 Why?
 "Because that's what the grammar says" isn't a valid answer.
 The C language is one of the most quirky ones around full of apparently ridiculous things. Why shouldn't you be able to write this for example:
     {
    ....
    L:
    }
 This stupid rule means that EVERY label in my generated code needs to be written as L:; instead of just L:
 Please don't say the label is only defined to be a prefix to another statement. I asking why it was done like that.

- It's reasonable not to require a semicolon (which is a statement termination) to terminate a statement block, which is terminated by the closing brace. A semicolon would be redundant.
- For labels, I agree, but the deeper issue is not that; it's to have goto's when better constructs would be the fix.