From: "Aaron Bannert" <[EMAIL PROTECTED]>
Sent: Saturday, July 21, 2001 1:52 AM
> [Still trying to get the hang of this whole patch-submission/OSS project/
> mailing list thing :) -- this patch should be *much* smaller]
Much improved :)
> Update to APR thread API to explicitly pass the apr_thread_t to the
> worker function.
Good choice
> Changed the worker routine's signature to take a single parameter:
> apr_thread_param_t, which contains the opaque data and the apr_thread_t.
Can we simply hide the details, and use our own _thread_main starting point?
Put the thread start function ptr in this structure as well, have our
_thread_main unwrap whatever is required (and it can create thread-local
data, if we ever implement that), finally calling out to the user's *func?
> httpd-2.0 will have to be updated after applying this patch to APR.
Not if we hide this detail, I believe. I don't know what that does about
providing access to the apr_thread_t in the child thread. Perhaps we declare
the user's func as accepting arguments (apr_thread_t *me, void *mine).
Bill
> Index: srclib/apr/include/apr_thread_proc.h
> ===================================================================
> RCS file: /home/cvspublic/apr/include/apr_thread_proc.h,v
> retrieving revision 1.65
> diff -u -r1.65 apr_thread_proc.h
> --- srclib/apr/include/apr_thread_proc.h 2001/06/06 18:11:06 1.65
> +++ srclib/apr/include/apr_thread_proc.h 2001/07/21 06:43:58
> @@ -125,7 +125,13 @@
> typedef struct apr_other_child_rec_t apr_other_child_rec_t;
> #endif /* APR_HAS_OTHER_CHILD */
>
> -typedef void *(APR_THREAD_FUNC *apr_thread_start_t)(void *);
> +struct apr_thread_param_t {
> + apr_thread_t *t;
> + void *data;
> +};
> +typedef struct apr_thread_param_t apr_thread_param_t;
> +
> +typedef void *(APR_THREAD_FUNC *apr_thread_start_t)(apr_thread_param_t *);
My only objection is making this new struct 'public'. If we 'actually' invoked
our own static start_thread function
