Add vector arithmetic #5149
Add vector arithmetic #5149
Conversation
Implement addition, subtraction, multiplication, division, and modulo for numeric (unsigned integer, signed interger, and floating point) vectors as well as addition for string vectors. We haven't yet worked out a way to write concise tests for vector operations so test coverage is minimal for now.
| } | ||
| for lform := vector.Form(0); lform < 3; lform++ { | ||
| for rform := vector.Form(0); rform < 3; rform++ { | ||
| name := "arith" + opToAlpha[op] + typ + lform.String() + rform.String() |
There was a problem hiding this comment.
Your naming convention differs from your compare pr in that the generated functions are seperated by underscores (e.g., cmp_EQ_Int_Flat_Flat). I think you should pick one for consistency.
There was a problem hiding this comment.
I'm planning to remove the underscores from gencomparefuncs.go in a separate PR.
| } | ||
|
|
||
| func (i *Int) Append(v int64) { | ||
| i.Values = append(i.Values, v) |
There was a problem hiding this comment.
Do nulls need to be append to here?
There was a problem hiding this comment.
No need. The model for vector operations whose output is a fixed-size primitive type (like Int, Uint, Float, and Bool) is that we compute a value for every slot in the output vector regardless of nulls in the input vectors and separately compute an output nulls vector (the Nulls *Bool field present in most of the vector types) from the input nulls vectors.
This PR doesn't include that nulls handling and neither does #5146 or any of the existing code under runtime/vam/expr. But it's coming.
Implement addition, subtraction, multiplication, division, and modulo for numeric (unsigned integer, signed interger, and floating point) vectors as well as addition for string vectors.
We haven't yet worked out a way to write concise tests for vector operations so test coverage is minimal for now.