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
Currently the VarZeroVec allocation is structured as [length, ...offsets, ...data]. However, there are reasons the length could come in externally, perhaps the length is already known because it is a map (#5376), or the length is known constant time (#5240).
We should have a VarZeroSliceUnbound type that represents [...offsets, ...data], whose functions require an externally-provided length. VZS will simply pass in its first length bytes, but other types can do fancier things.
(if not a separate type, we should at least refactor the VZV internal functions to be reusable this way)
Ideally the map polymorphism solution (#3128) will allow for a map type that externally stores a length as well. I'm not quite sure how that would work, but we can design that later.
The text was updated successfully, but these errors were encountered:
Currently the VarZeroVec allocation is structured as
[length, ...offsets, ...data]
. However, there are reasons the length could come in externally, perhaps the length is already known because it is a map (#5376), or the length is known constant time (#5240).We should have a VarZeroSliceUnbound type that represents
[...offsets, ...data]
, whose functions require an externally-provided length. VZS will simply pass in its first length bytes, but other types can do fancier things.(if not a separate type, we should at least refactor the VZV internal functions to be reusable this way)
Ideally the map polymorphism solution (#3128) will allow for a map type that externally stores a length as well. I'm not quite sure how that would work, but we can design that later.
The text was updated successfully, but these errors were encountered: