Replies: 5 comments 9 replies
-
|
I can't open the link but the idea sounds great! DuckDB and Julia are a great combo for DS. |
Beta Was this translation helpful? Give feedback.
-
|
The DuckDB and Julia Combo is amazing! Not sure about the links, they seem to work for me 🤔 I want to polish it a bit and create a draft PR |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your work! Links are working now, possibly my connection earlier. |
Beta Was this translation helpful? Give feedback.
-
|
Also can't see the work |
Beta Was this translation helpful? Give feedback.
-
|
@tqml Sorry I'm just seeing this. I just put in the following pr and it was merged. That being said your interface is much smoother .. I way prefer it. Is it possible for you to still add it ? We could possible figure out how to do aggregate functions too |
Beta Was this translation helpful? Give feedback.
-
|
Hey @drizk1, no worries I was quite busy last month and couldnt really continue working on it. Yeah, I definitely would like to add is as well :) My issue when I wrote it is, was that I couldn't get it to work with Strings (or structs), do you have any pointers there? I will polish up the code and create a draft/wip PR |
Beta Was this translation helpful? Give feedback.
-
|
The macro wouldn't work for strings or structs ? Or the scalar function? It's tough to say without being able to see the code. Would you be able to share it somehow? And I can try it out |
Beta Was this translation helpful? Give feedback.
-
|
The implementation currently produces a segmentation fault when one of the arguments has a data type other than Int/Float, which is most likely due to missing string casting. In my understanding, the vector in the data chunk is a vector of (smart) string pointers that would need conversion to the Julia native String type to be used in Julia (and of course also the other way round if the result should be of type String). |
Beta Was this translation helpful? Give feedback.
-
|
Oh I see for the one that was recently merged? |
Beta Was this translation helpful? Give feedback.
-
|
A quick syntax comment, it seems to me that fun = @create_scalar_function "my_sum" [Int64, Int64] Int64 my_sumCould be alternatively expressed as fun = @create_scalar_function mysum(::Int64, ::Int64)::Int64 as "my_sum"which reads a bit better to me. |
Beta Was this translation helpful? Give feedback.
-
|
Agreed, that reads a bit more intuitively 👍 |
Beta Was this translation helpful? Give feedback.
-
|
You could probably even drop the quotes in fun = @create_scalar_function my_sum(::Int64, ::Int64)::Int64 |
Beta Was this translation helpful? Give feedback.
-
|
This feature with the proposed interface was merged in PR #15430 :) |
Beta Was this translation helpful? Give feedback.
-
Hey!
I think this feature is pretty cool (as DuckDB in general) and since I have some extra time in the next two months I would like to help out with the Julia implementation of this feature.
I envision the Julia API to look like the following:
I created a first example here scalar_function.jl. It is based on the existing implementation of table functions with the addition to have a macro that can implement a specialized wrapper function the correct data types. The implementation requires the most recent build of duckdb (#4874 (comment)) but seems to work so far.
Let me know what you think!
Beta Was this translation helpful? Give feedback.
All reactions