Table like clause issue

[lelit]

Hi Lukas,

while working on my pglast deparser I noticed that the following statement

CREATE TABLE like_constraint_rename_cache   (LIKE constraint_rename_cache INCLUDING ALL)

gets deparsed by current libpg_query as

CREATE TABLE like_constraint_rename_cache (LIKE constraint_rename_cache INCLUDING ALL INCLUDING COMMENTS INCLUDING CONSTRAINTS INCLUDING DEFAULTS INCLUDING IDENTITY INCLUDING GENERATED INCLUDING INDEXES INCLUDING STATISTICS INCLUDING STORAGE)

It seems redundant, isn't it? If so, it this PR reasonable?

Unfortunately I could not run the tests locally, because a make stopped with

...
compiling src/postgres/src_port_strerror.c
compiling src/postgres/src_port_strnlen.c
compiling vendor/protobuf-c/protobuf-c.c
compiling vendor/xxhash/xxhash.c
Warning: protoc-gen-c not found, skipping protocol buffer regeneration
compiling protobuf/pg_query.pb-c.c
ar: Cannot convert existing thin library libpg_query.a to normal format
make: *** [Makefile:145: libpg_query.a] Errore 1

and I could not figure out what's wrong here... I'm on a Debian sid, if that matters.

[lfittl]

@lelit Nice, good catch!

I wonder if we should add this as an explicit test to the list in test/deparse_tests.c? (those are checked for matching 1:1 in the text - the regression tests are only checked for matching in the parse tree, but not in the exact text representation)

(I'll review the issue you've reported on running the tests separately - need to boot a VM with Debian sid to confirm)

[lelit]

Thanks!

[lfittl]

@lelit I've just ran make successfully (including the tests) on Debian sid, including running all the tests.

Could it be that you have a leftover libpg_query.a library from an older version of libpg_query?