I'd like to learn your thoughts on ordinal numbers in step IDs. They are generated in every step, so that if one wants to use the result of a given step, the step ID is already there.
They work fine, but in the diffs when we add/remove a step somewhere in the middle of the job, the IDs after the new/old step are changed. Are you generally fine with this? Do your workflows contain lots of steps so this behavior causes excessive pain?
I thought about changing the approach here: instead of adding an ordinal number, calculate a short hash form the step's command/action params, then the hash would remain unchanged as long as there step itself doesn't change. So it would look like
step-fd734cab
. Hash collisions would be caught in script's runtime and resolved somehow.
Let me know!