I've noticed that Disque, similar to Beanstalkd, returns a job id when you add a job, and seemingly does not provide a way for the client to supply the job id. Out of curiosity, is there any reason Disque must work this way?
The use case I'm interested in here is that applications may have well-known ID schemas, so they can cancel delayed jobs easily. For instance, if a new user signs up, we may want to send them an "Are you stuck?" email if they don't do some particular action within some amount of time. However, once they do that action, the application will want to cancel a job with the id are_you_stuck_<user_id>. With Disque returning job ids rather than accepting them, you'd have to maintain a mapping of application_id -> disque_id somewhere to make this scheme work. (Of course, there are ways to accomplish this particular task using other approaches.)
I've noticed that Disque, similar to Beanstalkd, returns a job id when you add a job, and seemingly does not provide a way for the client to supply the job id. Out of curiosity, is there any reason Disque must work this way?
The use case I'm interested in here is that applications may have well-known ID schemas, so they can cancel delayed jobs easily. For instance, if a new user signs up, we may want to send them an "Are you stuck?" email if they don't do some particular action within some amount of time. However, once they do that action, the application will want to cancel a job with the id
are_you_stuck_<user_id>. With Disque returning job ids rather than accepting them, you'd have to maintain a mapping ofapplication_id -> disque_idsomewhere to make this scheme work. (Of course, there are ways to accomplish this particular task using other approaches.)