Use anonymous struct/enum in typedefs where possible

[nrwahl2]

Very low priority, but should be a quick review

[clumens]

It would also be worth adding something to doc/sphinx/Pacemaker_Development/c.rst explaining this. I'm hoping one day we'll have someone new on the team and we can point them at that instead of having to re-explain all this stuff.

[clumens]

I would love to deprecate this all the way out of the code. It's so obnoxious to have a reimplementation of a linked list as public API.

[nrwahl2]

Agreed. That would be quite a bit more involved to do right now though. We have not only lrmd_list_freeall() using it but also three methods in lrmd_api_operations_t, which is not deprecated. Those method signatures can't be changed, so we'd have to get to a point where we can deprecate them altogether.

[clumens]

Is there a reason we can't also deprecate stonith_history_s and stonith_key_value_s? (Speaking of things I'd love to see go away)

[nrwahl2]

Yep! The reason is they're already deprecated.

/*!
 * \brief Key-value pair list node
 * \deprecated Do not use
 */
typedef struct stonith_key_value_s {
    char *key;
    char *value;
    struct stonith_key_value_s *next;
} stonith_key_value_t;

// @TODO Keep this definition but make it internal
/*!
 * \brief Fencing history entry
 * \deprecated Do not use
 */
typedef struct stonith_history_s {
    char *target;
    char *action;
    char *origin;
    char *delegate;
    char *client;
    int state;
    time_t completed;
    struct stonith_history_s *next;
    long completed_nsec;
    char *exit_reason;
} stonith_history_t;

[nrwahl2]

It would also be worth adding something to doc/sphinx/Pacemaker_Development/c.rst explaining this. I'm hoping one day we'll have someone new on the team and we can point them at that instead of having to re-explain all this stuff.

The development guide already says enums shouldn't have a typedef at all, and that they don't require any special naming convention. So really we should have enum crm_exit or enum crm_exit_code rather than enum crm_exit_e or crm_exit_t. Though I didn't feel like going that far, since we use crm_exit_t in tons of places internally and in function signatures.

I'll add a note for structs.