MAX() function always returns 0

[Matthieu-LAURENT39]

On 0.8.5, as well as on master (9d165b2), the MAX() function seems to always return 0 for anything related to dates.
Here are some queries i tried that have this issue:

• fselect "MAX(modified) from /tmp/test"
• fselect "MAX(YEAR(modified)) from /tmp/test"

However, this works:

• fselect "MAX(size) from /tmp/test"

Also note that

• fselect "YEAR(modified) from /tmp/test"
• fselect "modified from /tmp/test"
  work just fine

i'm currently trying to find the issue, i'll open a PR if i do find it

[Matthieu-LAURENT39]

Well, i found the issues.

The MAX function only works if it can cast the value to usize, otherwise it ignores the value, so that explains why fselect "MAX(modified) from /tmp/test" didn't work. I assume that's intended behavior.

The issue with fselect "MAX(YEAR(modified)) from /tmp/test" (as well as fselect "MIN(YEAR(modified)) from /home/user", which is an example from the docs) is that it will look for Year(Modified) in the hasmap from the raw_output_buffer, but that hasmap only contains the raw Modified and an empty Min(Year(Modified)).
That means a really simple workaround is to use this query: MIN(YEAR(modified)), YEAR(modified) from /tmp/test., which will make fselect compute the needed Year(Modified).

I tried fixing it myself, but i can't really understand the structure of the code, especially since it doesn't have many comments.

[jhspetersson]

Thank you very much for this issue!

Indeed, the code is an undocumented mess... It appears, a deliberate "shortcut" has been made at some point, so the aggregates didn't work with the enclosed functions.

I added a small fix to the master branch. It would be great to check if it works for you now.

MIN and MAX for datetime fields is actually an another problem. Will add it soon, I guess :)

[Matthieu-LAURENT39]

Works fine now!
About the lack of code comments, would you accept a PR that reformats / documents some of the code?

[jhspetersson]

Absolutely! That would be great, thank you.

…

On Sat, May 11, 2024, 11:22 Matthieu LAURENT ***@***.***> wrote: Works fine now! About the lack of code comments, would you accept a PR that reformats / documents some of the code? — Reply to this email directly, view it on GitHub <#148 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABOACPASJSBCR7LFOXY5AVTZBXPOFAVCNFSM6AAAAABHQ6U7AWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMBVGY2TANZTHE> . You are receiving this because you commented.Message ID: ***@***.***>