You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many Specter navigators do not preserve metadata on transforms, such as ALL, MAP-VALS, and MAP-KEYS. Should look into the performance cost of metadata preservation and whether it's worth updating all the navigators to do so. Implementing this would also eliminate the need for ALL-WITH-META, though it could be kept around for backwards compatibility as (def ALL-WITH-META ALL).
The text was updated successfully, but these errors were encountered:
I know performance was always a requirement with this library so I'm not surprised to see an "opt-in" for something that might have a performance cost like ALL-WITH-META versus other libraries and languages that have an "opt-in" for things that have performance benefits like type hinting, hand rolled in-lining, parallel versions of functions, etc
So I imagine if you wanted to go down that road you could provide an ALL-WITHOUT-META for those that are paranoid that hanging on to the metadata is costing nano-seconds.
Like I said, I'm aware that this library's promise has always been performance by default so I imagine you've already thought of this and rejected it.
Many Specter navigators do not preserve metadata on transforms, such as
ALL
,MAP-VALS
, andMAP-KEYS
. Should look into the performance cost of metadata preservation and whether it's worth updating all the navigators to do so. Implementing this would also eliminate the need forALL-WITH-META
, though it could be kept around for backwards compatibility as(def ALL-WITH-META ALL)
.The text was updated successfully, but these errors were encountered: