User Details
- User Since
- Mar 20 2019, 8:25 PM (96 w, 4 d)
- Roles
- Administrator
Fri, Jan 22
It would be nice to have a better story for these named tuples. Unfortunately, it might require mucking around with metaclasses to support overriding init if we want to do runtime type checking.
others have more context on this than I
mostly just concerned about leaving stuff around we don't need - lets not add the execution plan snapshot if we dont need it
back to you to parameterize dict/list
Thu, Jan 21
solids still have access to context.pipeline_def right? just making sure this isnt a breaking change
should be good after the ... changes
Do you have something in mind for how we could improve the display?
Tue, Jan 19
first reactions:
- concerns around hiding errors / confusion for intermediate user getting Any when trying to actually use types
- throwing away information by dropping all the way to Any
what led away from the approach in https://dagster.phacility.com/D5784 ? Just updating test call sites?
I think I'm ok with this - don't feel that strong in either direction. Curious for others thoughts.
Fri, Jan 15
nice thanks for the close review, will address inlines
Thu, Jan 14
does the watch thread need to get turned back on?
to your queue for mentioned test and additional comments
nice
zoom debugging / review
Wed, Jan 13
the current status is having one of these date yaml entries in a preset file like
^ flake
+ @prha for lint shenanigans
Mechanically this is a pretty clever way to get this feature working in the sharded sqlite scenario but the implementation is some real "inheritance is the root of all evil" shit. I think it makes it clear that the way the run sharding was managed between the child/parent class is pretty fragile foundation.
Tue, Jan 12
I do think it would be good to start stamping postgres alembic revisions, just so we have the flexibility to know for sure what revision a given user is on
i dont know, I think building missing indexes on migration seems reasonable, the tracking of indexes in each storage also makes sense. I think dropping all this stuff will have to come with some big change to the instance set up
i don't love this complexity burden but i think its the right call for end user experience
lets go for the consolidation while we're making changes, should just be easy enough
some complex stuff but i think the comments and cleanup make it parseable at least, thanks for that.