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
As commented in #143 (comment), Fixed/Normed (i.e. the difference in scaling factor) and signed/unsigned are orthogonal concepts.
Type (scaling)
Fixed (2^f)
Normed (2^f-1)
signed
✔️
❌ N/A
unsigned
❌ N/A
✔️
Historically, Fixed is signed and Normed is unsigned, but I think we should separate the two orthogonal concepts as much as possible.
I do not intend to support unsigned Fixed and signed Normed right now. However, I think that it is "safe" to describe in README.md that they may be supported in the future, because we have no systematic way to notify the downstream packages of the breaking changes to the existing Fixed and Normed.
(Whether relaxing is considered "safe" or "unsafe" depends on the point of view.)
At least I think it is worth writing that in CONTRIBUTING.md. It clarifies whether the code should be written in "src/FixedPointNumbers.jl", "fixed.jl" or "normed.jl". I think that the code which is not related to scaling factors should be written in "src/FixedPointNumbers.jl", i.e. should be commonized.
The text was updated successfully, but these errors were encountered:
As commented in #143 (comment),
Fixed
/Normed
(i.e. the difference in scaling factor) and signed/unsigned are orthogonal concepts.Fixed
(2^f
)Normed
(2^f-1
)Historically,
Fixed
is signed andNormed
is unsigned, but I think we should separate the two orthogonal concepts as much as possible.I do not intend to support unsigned
Fixed
and signedNormed
right now. However, I think that it is "safe" to describe in README.md that they may be supported in the future, because we have no systematic way to notify the downstream packages of the breaking changes to the existingFixed
andNormed
.(Whether relaxing is considered "safe" or "unsafe" depends on the point of view.)
At least I think it is worth writing that in CONTRIBUTING.md. It clarifies whether the code should be written in "src/FixedPointNumbers.jl", "fixed.jl" or "normed.jl". I think that the code which is not related to scaling factors should be written in "src/FixedPointNumbers.jl", i.e. should be commonized.
The text was updated successfully, but these errors were encountered: