Repository navigation
3 i parses as 3 ^ i: an imaginary literal after a factor is read as its exponent #1503
Description
Activity
The bigger question is whether we want juxtaposition to be interpretable as exponentiation at all; think about if
x 2should be read differently fromx(2).I'd make juxtaposition a product everywhere, so
x 2,x(2),(x + 1) 2,3 2and3 iwould all be products. As far as I know that's also how Mathematica and SymPy's implicit multiplication read a space.The power rule only earns its place in
x2, a digit written straight against a name. There it's genuinely ambiguous: it could meanx^2,x_2(the index form the lexer already has, since a name can only continue past its letters with_), orx * 2. I'd refuse it with a parse error that names those three, rather than choose one silently. Todayx2isx ^ 2andx1isx ^ 1. Movingx2tox * 2would silently change what an existing expression means, and a refusal can't do that.So:
now proposed x 2,(x + 1)2,3 2powers products 3 i,x i3 ^ i,x ^ i3 * i,x * ix2,a1x ^ 2,a ^ 1a parse error naming x^2,x_2andx*22x,x y,x(2)products unchanged If you'd rather keep
x2as a square for compatibility, the rest stands on its own: a number after a space, a bracket, or another number is a product.Queue this for v3 then for the breakage.
On the 3.0 milestone. When it is taken up, juxtaposition becomes a product everywhere, with a
BREAKING-CHANGES.mdentry for each reading that changes (x 2,3 2,3 i).
3 iparses as3 ^ i, not3 * i, sosin(3 i)is evaluated at3^i(about0.455 + 0.891i):2.5.0 does the same. The juxtaposition rule inserts
^before a number written after a number, a name or), so thatx2isx ^ 2. The lexer readsias a number, so the rule applies to it as well.Syntax.mdrecords this as a consequence of the rule since #1041, andSyntaxDocumentedTestpins"x i"as"x ^ i".Proposal: an imaginary literal written after a factor is a factor too.
3 iwould be3 * iandx 2iwould bex * 2i, whilex2,3 2and(x + 1)2stay powers.3i, written as one literal, is already3i. I'd change the rule and the documented row together. Is that the reading you want?