On 2/5/2008 3:48 PM, David Rees wrote:
> On Feb 5, 2008 6:57 AM, Brad Nicholson <bnichols at ca.afilias.info> wrote:
>> Builds, but doesn't work is my understanding.  Something to do with a
>> change of the number of arguments passed to an internal PG function.
> 
> Doesn't even build for me on my Fedora 8 test system against 8.3:
> 
> make[2]: Entering directory `/tmp/slony1-1.2.12/src/backend'
> gcc -g -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -I../..
> -fpic -I/usr/local/pgsql-8.3.0-2008020400/include/
> -I/usr/local/pgsql-8.3.0-2008020400/include/server/  -c -o
> slony1_funcs.o slony1_funcs.c
> slony1_funcs.c: In function 'slon_quote_literal':
> slony1_funcs.c:1106: warning: pointer targets in passing argument 1 of
> 'pg_mblen' differ in signedness
> slony1_funcs.c: In function 'getClusterStatus':
> slony1_funcs.c:1357: warning: passing argument 1 of 'typenameTypeId'

The problem is that 8.3's typenameTypeId() has changed to 3 call 
arguments. The config/acx_libpq.m4 of REL_1_2_STABLE does check for that 
correctly. The relevant change was committed on November 22., which was 
after 1.2.12 was released.

I kinda disagree that we should have released 1.2.13 BEFORE 8.3 was 
actually out. As long as something is a release candidate, there is 
still a faint chance that some API may change (however small that might 
be). So what people are actually asking for is to release with the first 
RC, and rerelease with every subsequent RC that might break anything in 
Slony. I don't think we have the resources to do that.


Jan


> Pulling the HEAD of CVS does build as does the REL_1_2_STABLE branch,
> but I have not installed it or ran any tests.
> 
> -Dave

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #