> OK, well, I applied pgsql-v9.2-drop-reworks-2.v4.1.patch and tried to
> compile, and got this:
> In file included from ../../../src/include/catalog/dependency.h:17,
> from dependency.c:19:
> ../../../src/include/catalog/objectaddress.h:21: warning: type
> defaults to ‘int’ in declaration of ‘ObjectAddress’
> ../../../src/include/catalog/objectaddress.h:21: error: expected ‘;’,
> ‘,’ or ‘)’ before ‘*’ token
> The problem here is pretty obvious: you've defined
> get_object_namespace, which takes an argument of type ObjectAddress,
> before defining the ObjectAddress datatype, which is the next thing in
> the same header file. How does that even compile for you?
Sorry, I didn't write out dependency of the patches.
Please apply the patches in order of part-1, part-2 then part-3.
I checked correctness of the part-2 on the tree with the part-1 already.
Both of the part-1 and part-2 patches try to modify objectaddress.h,
and the part-2 tries to add get_object_namespace() definition around
the code added by the part-1, so the patch commands get confused
and moved the hunk in front of the definition of ObjectAddress.
[kaigai(at)iwashi pgsql]$ cat
~/patch/pgsql-v9.2-drop-reworks-2.v4.1.patch | patch -p1
patching file src/backend/catalog/objectaddress.c
Hunk #1 succeeded at 976 (offset -429 lines).
patching file src/include/catalog/objectaddress.h
Hunk #1 succeeded at 17 with fuzz 2 (offset -18 lines).
patching file src/test/regress/expected/drop_if_exists.out
I'm sorry again. I tought it was obvious from the filenames.
* Part-2 (depends on Part-1)
* Part-3 (depends on Part-1 and -2)
KaiGai Kohei <kaigai(at)kaigai(dot)gr(dot)jp>
In response to
pgsql-hackers by date
|Next:||From: Thom Brown||Date: 2011-10-10 17:39:20|
|Subject: Re: Range Types - typo + NULL string constructor|
|Previous:||From: Kevin Grittner||Date: 2011-10-10 17:36:26|
|Subject: Re: COUNT(*) and index-only scans|