-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 16/11/13 17:46, Schmurfy wrote: > I definetely like the idea of having the existing constructs > supporting async use, I used both zeromq and crossroads in > different personal projects but in the idea I only used XREQ/XREP > since I always had multiple req/rep in parallel. As for the > request_id the user should have full control over it but why not > add a new function in the library to generate ids in sequence ? > > int nn_init_sequence(int *seq); int nn_next_id(const int *seq); Another option would be to use user-provided IDs. Specifically void pointers would be nice as it allows to associate the request with arbitrary user-defined object. nn_req_send (int s, void *id, void *buf, size_t len, int flags); > Like paul I keep implementing and reimplementing the same stuff > again and again, having that builting would really be a nice > thing. Yes. It looks like the parallel requests happen in more than 50% of use cases, so implementing it in the library makes sense. Martin -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQEcBAEBAgAGBQJSh78YAAoJENTpVjxCNN9Y1JQH/R+s5fEfnAEduW73L7QqUOLK 4+kaOpBoomGLc/DsLt+ZPMqZThZKK3gI6r+1L4M7Bt7IONDs+wo1gy9eqmpnoL6J 0hO8h/gN23pFcvBwiA6Udf6c7lAk8QGF4tLycrxVcCP8yBIPBDLys18rGqgpbcHu F3AKcKiCCChFg+BNXO7M4v0NcegS1tcvHnAAVqpFh3si6GHI9x80atlwuM+84SeI tb4K+YiOZHNwmgCXNu9mYMMLXHjD1f008vEHHsnowpOIKISI3q24mooE6YJq6bWG 7BDoEp7Y+Vq2iE283CdtIkGim9nEBbA87mXBz/QhqHNVqTVihB3EG/UVchiGyEE= =sqNJ -----END PGP SIGNATURE-----