# Implementing Tables.jl interface results in Stack Overflow

**URL:** https://discourse.julialang.org/t/implementing-tables-jl-interface-results-in-stack-overflow/102633
**Category:** General Usage
**Created:** [August 9, 2023, 11:58am UTC](https://discourse.julialang.org/t/implementing-tables-jl-interface-results-in-stack-overflow/102633 "2023-08-09T11:58:19Z")
**Posts on this page:** 2
**Page:** 1

<div class="post-metadata">

### Author: ![MitchellWeg](https://avatars.discourse-cdn.com/v4/letter/m/5fc32e/32.png) [@MitchellWeg](https://discourse.julialang.org/u/MitchellWeg)
#### Post date: [August 9, 2023, 11:58am UTC](https://discourse.julialang.org/t/implementing-tables-jl-interface-results-in-stack-overflow/102633/1 "2023-08-09T11:58:19Z")

</div>

For my package, I’m trying to implement the Tables.jl package. I however don’t really have a clue on how to implement the interface, and what the specific functions need to return.

Currently, I have this code. It however results in a stack overflow:

```julia
mutable struct TextCursor <: DBInterface.Cursor
  conn::DB
  sql::String
  nfields::Int
  nrows::Int
  rows_affected::Int64
  result::Vector{Vector{internal_column}}
  names::Vector{Symbol}
  types::Vector{Type}
  lookup::Dict{Symbol,Int}
  current_rownumber::Int
  current_resultsetnumber::Int

  function TextCursor(db, sql, data)
    nfields = length(data) > 0 ? length(data[1]) : 0
    nrows = length(data)
    rows_affected = 0
    current_rownumber = 0
    current_resultsetnumber = 0
    types = length(data) > 0 ? get_types(data) : Vector()
    names = length(data) > 0 ? get_names(data) : Vector()
    lookup = length(data) > 0 ? Dict(x => i for (i, x) in enumerate(names)) : Dict()

    cursor = new(db, sql, nfields, nrows, rows_affected, data, names, types, lookup, current_rownumber, current_resultsetnumber)

    return cursor
  end

  function get_types(data)
    types = Vector()

    for col in data[1]
      _type = col.type
      push!(types, _type)
    end

    return types
  end

  function get_names(data)
    names = Vector()

    for col in data[1]
      symbol_name = Symbol(col.name)
      push!(names, symbol_name)
    end

    return names
  end
end

struct TextRow <: Tables.AbstractRow
  row::Int
  cursor::TextCursor
end

Base.eltype(c::TextCursor) = internal_column
Tables.istable(::Type{TextCursor}) = true
Tables.rowaccess(::Type{TextCursor}) = true
Tables.rows(c::TextCursor) = c
Tables.columnnames(r::TextRow) = r.cursor.names

function Tables.getcolumn(r::TextRow, i::Int)
  println(r)
end

 function Tables.getcolumn(r::TextRow, nm::Symbol)
  println(r)
 end

Base.iterate(c::TextCursor, st=1) = st > c.nrows ? nothing : (TextRow(st, c), st + 1)

```

Within the getcolumn functions, the stack overflow happens. For some odd reason, it keeps printing the row a million times and crashes. I was wondering, what does the specific function need to return?

---

<div class="post-metadata">

### Author: ![algunion](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/algunion/32/51630_2.png) [@algunion](https://discourse.julialang.org/u/algunion)
#### Post date: [August 10, 2023, 12:57am UTC](https://discourse.julialang.org/t/implementing-tables-jl-interface-results-in-stack-overflow/102633/2 "2023-08-10T00:57:19Z")

</div>

> [@MitchellWeg](#):
>
> I however don’t really have a clue on how to implement the interface, and what the specific functions need to return.

Please use the following as a general guideline: [Implementing the Interface (i.e. becoming a Tables.jl source)](https://tables.juliadata.org/stable/#Implementing-the-Interface-(i.e.-becoming-a-Tables.jl-source))

Please take a close look at the `required` methods and their stated return value compatibility: for example, `Tables.rows` needs to return an `Tables.AbstractRow`-compatible iterator.

Another, maybe more practical, approach is to dive into some of the [existing integrations](https://github.com/JuliaData/Tables.jl/blob/main/INTEGRATIONS.md). Besides seeing how others successfully implemented the interface, you might even find some that are pretty close to your data structure.

I hope this gets you started - if there are still things that cause issues after applying the methodology from above, please follow up on this topic so we can get into those.