Re: Small issues with CREATE TABLE COMPRESSION

On Sat, May 08, 2021 at 08:19:03PM +0530, Dilip Kumar wrote:
> I have fixed some of them, I could not write the test cases where we
> have to ensure that 'ALTER TABLE COMPRESSION' is not appearing in the
> dump. Basically, I don't have knowledge of the perl language so even
> after trying for some time I could not write those 2 test cases. I
> have fixed the remaining comments.

Thanks. I have spent some time on that, and after adding some tests
with --no-toast-compression, I have applied this part.

Now comes the last part of the thread: support for the build with
MSVC. I have looked in details at the binaries provided by upstream
on its release page, but these are for msys and mingw, so MSVC won't
work with that.

Saying that, the upstream code can be compiled across various MSVC
versions, with 2010 being the oldest version supported, even if there
is no compiled libraries provided on the release pages. The way of
doing things here is to compile the code by yourself after downloading
the source tarball, with liblz4.lib and liblz4.dll being the generated
bits interesting for Postgres, so using
https://github.com/lz4/lz4/releases as reference for the download
looks enough, still that requires some efforts from the users to be
able to do that. Another trick is to use vcpkg, but the deliverables
generated are named lz4.{dll,lib} which is inconsistent with the
upstream naming liblz4.{dll,lib} (see Makefile.inc for the details).
My image of the whole thing is that this finishes by being a pain,
still that's possible, but that's similar with my experience with any
other dependencies.

I have been spending some time playing with the builds and that was
working nicely. Please note that you have missed an update in
config_default.pl and not all the CFLAGS entries were present in
GenerateFiles().

It may be nice to see if this stuff requires any adjustments for msys
and mingw, but I don't have such environments at hand.

All that leads me to the updated version attached.

Thoughts?
--
Michael