Sorting when merging adjacent objects?

[javafanboy]

Assuming I have sorted my data, possibly using several columns, in some suitable order to work well with the min/max information for each column on row group level how can I control that this ordering is maintained when adjacent objects are merged?

I am refering to situations where just aggregating the smaller files would not give the desired result - assume a table sored on a composite of the two first columns (firstly column 1 then 2):

AAAA 123
AAAA 124
BBBB 100
BBBB 101
.................
AAAA 125
BBBB 200

Just aggregating would result in:
AAAA 123
AAAA 124
BBBB 100
BBBB 101
AAAA 125
BBBB 200

while desired result would be maintaining the desired sorting criteria of first column 1 and then column 2:
AAAA 123
AAAA 124
AAAA 125
BBBB 100
BBBB 101
BBBB 200

[dylanzal]

+1, this would be really helpful for my use case.

Merging non-adjacent files is currently supported (i.e., there are gaps in files' respective row ids) by adding a '_ducklake_internal_row_id' column to the written files. Therefore, it would seem that permitting sorting during compaction would be relatively straightforward? I've implemented this myself in a custom compaction utility in python and it achieves the desired effect. However, I'd much prefer to not rely on my hand-rolled implementation. Are there plans to support this in the near future?