Reijo Sund

Please see also #3229

Could you open a bit more what is the actual idea of this change? I guess that the purpose would be to reduce the number of dependencies of the actual...

Thanks for the reprex. Can't test it right now, but could you please also show the produced SQL by piping to `show_query()`. If the problem could then be repeated by...

Well, I suspect that this may be related more to `tbl_lazy()` mechanism than be an actual problem in the database. Therefore it is essential to see what `rows_update()` is doing...

OK, so it is permanently updating the original table in the database (instead of making changes only to temporary tables). That may not be desirable behaviour in many user cases,...

I was able to reproduce the bug. It seems to be a DuckDB issue as it happens even with pure SQL. I tried to minimize the reprex and it can...

OK, here is the reduced data and R script using only SQL queries. It is essential to start with a "fresh" copy of the data every time as the data...

Steps 1, 2 and 3 would be enough, especially the step 3 would be interesting as (the R script for) it is where the conditions for the problem are potentially...

Thanks for the data generation script. I was able to reduce it further to the following one (that still results in the bug most of the time on the latest...

> Is this possibly a good place to also look at #4116? It seems like the `blob` type can't be written from R, so possibly should be included among the...