Dimitri Fontaine

This looks like a `malloc()` problem and I believe by using a Garbage Collector as we do now we should be okay. Please try with the current sources of pgcopydb,...

See also #726. I am wondering if there is a way to inquire about privileges to edit a setting before running the ALTER DATABASE SET command, or maybe we should...

Closing this PR as after discussions, if I remember correctly, the solution looks like adding support to the pgoutput logical decoding plugin where it is possible to handle the filtering...

Is my understanding correct that the table (here `public.transaction`) is a plain table on the source database, but a partitionned-table (and the parent table) on the target database?

I think you need to split the schema and data parts of the processing, and then tweak the schema on the target database to introduce partitioning, in order to repro....

We have many cases to consider, but today only consider a few. The deal with TRUNCATE is how we use it in the same transaction as the COPY statement, in...

> I think the alternate approach can be accessing the catalog from ld_wal2json.c & ld_test_decoding.c, ignore the generated column while transforming to the data object. This sounds like it would...

> @dimitri It would be useful if can review this. Thank you. I should be able to review that early next week now. The work I'm doing to refactor the...

> @dimitri. Since we want exclude generated column for all INSERT & UPDATE, on the fly lookup into sqlite would add runtime overhead regardless of whether the source has any...

Hi @andres-chavez-bi ; I think this happens when there is no change happening on the source database anymore. Can you confirm it is also the issue you have?