[drmaa-wg] Wait Type

Andreas Haas Andreas.Haas at Sun.COM
Fri Feb 11 05:32:20 CST 2005


I support this proposal 100% !

Andreas

On Fri, 11 Feb 2005, Daniel Templeton wrote:

> Here's my proposal for the new wait structure:
>
> typedef struct drmaa_job_info_s drmaa_job_info_t;
>
> We then then add the functions:
>
> void drmaa_release_job_info(drmaa_job_info_t* info);
>
> int drmaa_wreason(char *reason, size_t reason_len, drmaa_job_info_t
> *info, char* error_diagnosis, size_t error_diag_len);
>
> int drmaa_wrusage(drmaa_attr_values_t **rusage, drmaa_job_info_t *info,
> char* error_diagnosis, size_t error_diag_len);
>
> int drmaa_wjobid(char *jobid, size_t jobid_len, drmaa_job_info_t *info,
> char* error_diagnosis, size_t error_diag_len);
>
> and change existing functions like this:
>
> int drmaa_wait(const char *job_id, signed long timeout, drmaa_job_info_t
> **info, char *error_diagnosis, size_t error_diag_len);
>
> int drmaa_wifexited(int *exited, drmaa_job_info_t *info, char*
> error_diagnosis, size_t error_diag_len);
>
> int drmaa_wexitstatus(int *exit_status, drmaa_job_info_t *info, char*
> error_diagnosis, size_t error_diag_len);
>
> int drmaa_wifsignaled(int *signaled, drmaa_job_info_t *info, char*
> error_diagnosis, size_t error_diag_len);
>
> int drmaa_wtermsig(char *signal, size_t signal_len, drmaa_job_info_t
> *info, char* error_diagnosis, size_t error_diag_len);
>
> int drmaa_wcoredump(int *core_dumped, drmaa_job_info_t *info, char*
> error_diagnosis, size_t error_diag_len);
>
> int drmaa_wifaborted(int *aborted, drmaa_job_info_t *info, char*
> error_diagnosis, size_t error_diag_len);
>
>
> I put the job_id_out and rusage into the new structure because that's
> how it's handled in the OO specs.  If the consensus is that we'd rather
> keep them as individual parameters, we can do that too.  I would argue
> that the job_id_out definitely belong in the new structure.  Barring
> what the OO specs do, I'm not sure how I feel about rusage.
> The error codes and function descriptions would obviously mirror those
> of the peer functions.
>
> Daniel
>
>





More information about the drmaa-wg mailing list