Remove unused ignored_subst_vars argument from Deploy._subst_vars#156
Remove unused ignored_subst_vars argument from Deploy._subst_vars#156rswarbrick wants to merge 1 commit intolowRISC:masterfrom
Conversation
Signed-off-by: Rupert Swarbrick <rswarbrick@lowrisc.org>
AlexJones0
left a comment
AlexJones0
left a comment
There was a problem hiding this comment.
I would need @machshev to weigh in, but I would hazard a guess that this was implemented with the long-term goal of staged evaluation of wildcards when merging the HJSONs? And so maybe it's worth keeping?
A current limitation of the system is that, if we were using --reseed 10000, we must make 10000 individual Deploy objects, and then evaluate the wildcards recursively for each of these 10000 objects. We need to do this because the HJSON could depend on the per-deploy unique seed values. But if we were more clever, we could substitute in all the wildcards except the Deploy-dependent ones (seeds, etc.), pass that to the Deploy objects, and then we only have to evaluate a couple of wildcards at most for each Deploy.
I believe that was the long term goal. At least, that's why it takes 5-10 minutes to even build the JobSpec graph to give the scheduler for an OpenTitan weekly run on my device 😅
2 participants
No description provided.