Re: exposing pg_controldata and pg_config as functions

On Tue, Feb 16, 2016 at 5:29 AM, Joe Conway <mail@joeconway.com> wrote:
> I believe this takes care of all open issues with this, so I plan to
> commit it as attached in a day or two. Thanks for your reviews and comments!

Here are just a couple of cosmetic comments.

+   The view <structname>pg_config</structname> describes the
+   compile-time configuration parameters of the currently installed
+   version of PostgreSQL. It is intended, for example, to be used by
+   software packages that want to interface to PostgreSQL to facilitate
+   finding the required header files and libraries. It provides the same
+   basic information as the <xref linkend="app-pgconfig"> PostgreSQL Client
+   Application. There is a System Information Function
Missing markup <productname></> around PostgreSQL.

+   Application. There is a System Information Function
Why is using upper-case characters necessary here? This could just say
system function.

The paragraph in func.sgml is a copy-paste of the one in
catalogs.sgml. We may want to remove the duplication.

+   /* let the caller know we're sending back a tuplestore */
+   rsinfo->returnMode = SFRM_Materialize;
I guess one can recognize your style here for SRF functions :)

@@ -61,7 +74,7 @@ libpgcommon_srv.a: $(OBJS_SRV)# a hack that might fail someday if there is a *_srv.o without a#
corresponding*.o, but it works for now.%_srv.o: %.c %.o
 
-   $(CC) $(CFLAGS) $(subst -DFRONTEND,, $(CPPFLAGS)) -c $< -o $@
+   $(CC) $(CFLAGS) $(subst -DFRONTEND ,, $(CPPFLAGS)) -c $< -o $@
Diff noise?

--- /dev/null
+++ b/src/common/config_info.c
[...]
+ * IDENTIFICATION
+ *   src/common/controldata_utils.c
This is incorrect.

+ * IDENTIFICATION
+ *   src/backend/utils/misc/pg_config.c
+ *
+ */
I am nitpicking here but this header block should have a long
"----------------" at its bottom.
-- 
Michael