On Jun 1, 2014, at 4:09 PM, Philip Guenther <[email protected]> wrote:
> On Sat, 31 May 2014, Brent Cook wrote:
>> with pqueue moving to a private interface, the typedef can occur twice
>>
>> ../include/openssl/dtls1.h:147:25: error: redefinition of typedef 'pqueue'
>> is a C11 feature [-Werror,-Wtypedef-redefinition]
>> typedef struct _pqueue *pqueue;
>> ^
>> ../include/pqueue.h:63:25: note: previous definition is here
>> typedef struct _pqueue *pqueue;
>
> It would be better to just eliminate the typedef from the public header,
> IMO. Since pqueue is a private interface and the header defining the
> struct isn't even exported, there's no (sane) way for someone else to use
> the typedef.
>
> (Side note: a #define like HAVE_STRUCT_PQUEUE not only falls in the
> application namespace, but in the namespace that autohell squats on, make
> it a bad choice for anything not related to autohell.)
Agreed - updated and will resend.
>
> Philip Guenther
>
>
> Index: dtls1.h
> ===================================================================
> RCS file: /cvs/src/lib/libssl/src/ssl/dtls1.h,v
> retrieving revision 1.11
> diff -u -p -r1.11 dtls1.h
> --- dtls1.h 30 May 2014 14:30:50 -0000 1.11
> +++ dtls1.h 1 Jun 2014 20:56:03 -0000
> @@ -144,11 +144,10 @@ struct dtls1_timeout_st {
> };
>
> struct _pqueue;
> -typedef struct _pqueue *pqueue;
>
> typedef struct record_pqueue_st {
> unsigned short epoch;
> - pqueue q;
> + struct _pqueue *q;
> } record_pqueue;
>
> typedef struct hm_fragment_st {
> @@ -191,10 +190,10 @@ typedef struct dtls1_state_st {
> record_pqueue processed_rcds;
>
> /* Buffered handshake messages */
> - pqueue buffered_messages;
> + struct _pqueue *buffered_messages;
>
> /* Buffered (sent) handshake records */
> - pqueue sent_messages;
> + struct _pqueue *sent_messages;
>
> /* Buffered application records.
> * Only for records between CCS and Finished
