Hi! I was wondering if the ids for LSST will have any maximum value, for example if we can say for certain that any diaSourceId will be smaller than 10^12 or something like that. I know from the precision of diaSourceId in the Alert Production Database and the Kafka schema of the alerts that this field will be a 64 bit integer, but I don’t know if the ids will be created incrementally or if they will take values all across the possible range for a 64 bit integer.
The reason I’m asking this is that I would like to create unique ids (for sources and objects) across many surveys/telescopes, which would be possible if LSST ids do not span the whole 64-bit integer range by using a survey prefix.
Cheers,
Hi Ignacio, thanks for this question. In this situation, it is recommended to assume that LSST ID values will span the whole range. A survey prefix, like you say, sounds like the best option.
I’m going to mark this post as the solution because I think it answers the question, but, please feel free to respond or open a new topic with new questions anytime.
Unfortunately, I suspect what Ignacio wanted is for his survey prefix to fit within the 64-bit integer. Instead, he will have to have a separate field to identify the survey. (To enable parallelization, such ids have many fields and are not strictly sequential counters. As a result, we’re reserving all 64 of the bits for our own use.)