Hi,
the coding standard for MySQL says the following about struct typedefing:
"Structure types are typedef'ed to an all-upper-case identifier"
However, I cannot find a single logical reason for this practice. Rather
I think it is a source for confusion, because typedefs are used rather
inconsistently within the code.
Examples:
typedef st_select_lex_node SELECT_LEX_NODE;
(st_ prefix is not carried on to the typedef).
struct TABLE_share;
(Mixed upper/lower case)
struct TABLE_SHARE;
(Uppercase - should be typedef according to coding style doc)
typedef struct st_field_info {...} ST_FIELD_INFO;
(typedef name is uppercased version of struct name).
typedef class Item COND;
(COND is an alias to Item - I guess this is allowed according to code
style).
struct TABLE_LIST;
(All uppercase letters)
typedef struct keyuse_t {...} KEYUSE;
(suffix _t on struct name)
typedef struct st_join_table {...} JOIN_TAB;
(st_join_table -> JOIN_TAB?)
class JOIN;
(typedef names are allowed to be uppercase - class names are not)
It also seems to me that C++ does not require typedef for shorthand
references to type names.
E.g the following compiles in gcc:
typedef class xxx_class {
int dummy;
};
struct XXX {
xxx_class *xxx;
yyy_class *yyy;
struct xxx_class *zzz;
class xxx_class *www;
};
The C++ book by Stroustrup says the following about typedefs:
"Use a typedef to define a meaningful name for a built-in type in cases
in which the built-in type used to represent a value might change", and
"Use typedefs to define synonyms for types" (implying the synonyms to be
a shorthand).
I propose that the guideline for typedef should be:
typedef can be used in the following cases:
- to define a meaningful name for a built-in type in cases in which
the built-in type used to represent a value might change
- to define a shorthand synonym for a complex type name
(e.g. typedef List<Table> Table_list)
- to define a synonym for another class when both names are meaningful:
(e.g. typedef class Item Condition)
This case should be used with great care to avoid confusion, however.
The coding standard should also include a note about legacy typedef
definitions:
Typedef definitions that do not follow this style should be removed when
natural cleanup of the code base occurs.
Thanks,
Roy
