Take advantage of UnitTuple to store type info #78
Conversation
|
Indeed, this package was written for 0.2, where that wan't possible yet. In particular #5, the earliest still open issue in this package. |
|
Oh, I didn't see that. Yea, that makes sense. Do you still want to support 0.2 with this package, or could something like this be merged? |
|
No, 0.2 is obsolete at this point
|
|
Ok. I noticed the latex output is broken on new version, and that the coverage tool is failing. Are you fine with change otherwise? |
| @@ -7,15 +7,28 @@ module SIUnits | |||
| import Base: ==, +, -, *, /, .+, .-, .*, ./, //, ^ | |||
| import Base: promote_rule, promote_type, convert, show, mod | |||
|
|
|||
| immutable SIQuantity{T<:Number,m,kg,s,A,K,mol,cd,rad,sr} <: Number | |||
| typealias UnitTuple NTuple{9,Int64} | |||
There was a problem hiding this comment.
Is Int64 really the correct thing here? Will this still be able to represent units like s^(1/2)?
There was a problem hiding this comment.
I haven't tried it with rational exponents, but there doesn't seem to be
anything that prevents it. Does it currently work properly with rational
exponents?
On Thu, Nov 19, 2015 at 11:22 PM Tomas Lycken notifications@github.com
wrote:
In src/SIUnits.jl
#78 (comment):@@ -7,15 +7,28 @@ module SIUnits
import Base: ==, +, -, , /, .+, .-, ., ./, //, ^
import Base: promote_rule, promote_type, convert, show, mod
- immutable SIQuantity{T<:Number,m,kg,s,A,K,mol,cd,rad,sr} <: Number
- typealias UnitTuple NTuple{9,Int64}
Is Int64 really the correct thing here? Will this still be able to
represent units like s^(1/2)?—
Reply to this email directly or view it on GitHub
https://github.com/Keno/SIUnits.jl/pull/78/files#r45441519.
There was a problem hiding this comment.
Actually, I was quite sure it should work (in fact, IIRC the move from ints-and-types to (values-of-any-bitstypes)-and-types as valid type parameters was made in base Julia with e.g. s^(1/2) as the main use case), but when I try now, it doesn't:
julia> using SIUnits; Second^(1/2)
1.0
Don't know if this is a regression or if it never worked, though.
There was a problem hiding this comment.
There is a test case that checks that sqrt(1s) throws an error, so I assume that hasn't worked for a while.
But with some minor changes, I was able to switch it to
typeaias UnitTuple NTuple{9,Rational{Int64}}It seems to work fine with some basic testing, and I would think this would be superset of the current functionality of this PR. Not sure if this is what you had in mind, or something more generic like
typeaias UnitTuple{N<:Number} NTuple{9,N}I would guess that this would require a decent amount of work to get working
I tried recreating this package as an exercise to understand the type system better. I found that using UnitTuple directly in SIUnit and SIQuantity simplified things quite a bit. I updated this package to work that way as well. Its a big change, but the tests are passing for me, so hopefully I fixed everything that needs to be fixed. I have a few other changes that simplify a few things, but this was the biggest change.