Re: PostgreSQL replication on Windows

Martha Stewart called it a Good Thing when xzilla(at)users(dot)sourceforge(dot)net (Robert Treat) wrote:
> On Sunday 23 January 2005 08:41, Christopher Browne wrote:
>> Martha Stewart called it a Good Thing when xzilla(at)users(dot)sourceforge(dot)net
> (Robert Treat) wrote:
>> > On Thursday 20 January 2005 06:14, Harm van der Ploeg wrote:
>> >> Are there any replication solutions that have been tested on
>> >> Windows. The options listed on the PostgreSQL site all require
>> >> Linux/Unix or are only available as source. I am not able to
>> >> compile these sources.
>> >
>> > I think all of the open source ones require *nix, although I
>> > think you could set up slony on a central machine and use it to
>> > replicate remote windows servers if you wanted.
>>
>> For Slony-I to function, for instance, requires having some
>> triggers implemented in C which would therefore have to be compiled
>> on Windows.
>
> Christopher, you know the code far better than I, but I don't think
> there is anything too crazy in the C triggers in slony that would
> make it difficult to compile on Windows (relative to other C
> functions anyways), do you ?

No, there's no reason to expect these triggers to be problematic.

What _is_ problematic is that the average Windows install doesn't
include a C compiler, making it rather tough for someone to add
anything in after the fact.

Then there's the challenge that if YOU compile the C functions with
YOUR environment, the binaries might not be easy to guarantee to match
up against a set of PostgreSQL binaries compiled by someone else.

If your compile environment matches well what is being used to produce
"quasi-official PostgreSQL binaries for Windows," then there's reason
for hope.

One rather less hopeful instance of this is that I _know_ that I can't
hook in Perl stuff compiled using GCC to the AIX 'standard Perl' (that
was presumably compiled using VisualAge C). GCC and VAC don't link
together happily enough for Perl stuff to "turn out." Porting trouble
found there...
--
"cbbrowne","@","gmail.com"
http://www3.sympatico.ca/cbbrowne/sap.html
It is better to have 100 functions operate on one data structure than
10 functions on 10 data structures. -- Alan Perlis