The convention here is to bottom post, or to intersperse comments, like in all the replies to you.
So it would be appreciated if you did that, rather than top post as you have been doing.
Thanks for the advice. I will follow the convention.
I strongly suspect that:
(1) making pg handle more than 1600 columns, would be way more complicated than you can imagine
(2) suich a change would be unlikely to be accepted into the main line, which would mean you'd have to reapply your patch for every new version of pg you wanted to use!
Yes. Our 1st attempt showed us (1).
Regarding to (2), We are good by adding a patch and recompile a patched version for our server databases.
But we are open on helping to add thousands of columns support as a compile-time parameter if there are other people interested.
Something like --with_thousands_of_columns_support feature
We could add a node to the build farm to test this functionality on Ubuntu. And helping and writing tests.