Opening up the Postgres codebase, I see that much of the C code is written by having pointers with the -> notation in such a way that:
(foo)->next = 5;
I know that pointer notation has levels of precedence, such that -> = (*foo). and is not the same as *foo.
However, does it mean anything when the parentheses are outside the variable name and de-referencing the address of next or is it merely a convention that is endemic to a coding style?
It is meaningless. The parenthesis have no effect since the expression inside them is so simple. It's basically
(*(foo)).next = 5;
Which, as you can see, is an extraneous set of parenthesis. If that's how the project is doing it, go for it. Otherwise, I think it looks rather awful, so I would try not to use it.
Its a coding convention that I've not seen before.
But it doesn't change anything.
(foo)->next = 5;
is exactly equivalent to
foo->next = 5;
It is a coding convention for safety.
In simple code, where foo is just a pointer, the parens are redundant.
(foo)->next = 5;
However, consider if there is or was the chance that foo might be defined by a macro so it is a more complex expression.
In that case, the parens might be required to avoid a compiler error or, worse, to avoid incorrect precedence. Imagine if it translated to an ugly expression like:
( ++get_foo() )->next = 5;
or a common cast:
( (db_record*)foo_ptr )->next = 5;
Something to bear in mind with idiomatic C code is that macros were often used as an abstraction mechanism for things we might do more explicitly in other languages ;-)