I apologize for assuming this was a well known problem.
If it's not feasible to adjust the build to work around this issue, I'd suggest the documentation
for pgAdmin be updated to clarify that it requires 8.3(?) or greater version of postgresql to build.
It's certainly not reasonable to assume people will be willing to remove their vendor provided and
supported versions of postgresql, but this documentation update would help people to know they need
to build a 1.10.0 version if they use Redhat 5. That's the rpm I build for my userbase and it mostly
works fine.
By the way, my users where specifically looking for the graphical query builder, which was my reason
for looking into version 1.10.x of pgAdmin.
dale
On 3/17/2010 11:33 AM, Dave Page wrote:
> 2010/3/17 Guillaume Lelarge<guillaume@lelarge.info>:
>> Le 17/03/2010 17:01, Devrim GÜNDÜZ a écrit :
>>> On Wed, 2010-03-17 at 16:52 +0100, Guillaume Lelarge wrote:
>>>> You should because I don't have any idea what you are talking about.
>>>
>>> I sent build log yesterday:
>>>
>>> http://archives.postgresql.org/message-id/1268719771.2229.1.camel@hp-laptop2.gunduz.org
>>>
>>
>> Oops, sorry. Didn't think they were related.
>>
>> I found this:
>>
>> pg_valid_server_encoding_id() is exported by 8.3's libpq, but it does
>> not exist in 8.2 and before. What is evidently happening is that psql
>> is binding to an old copy of libpq.so. Check your PG installation.
>>
>> in a Tom Lane's mail on the same kind of issue.
>>
>> Our issue comes from revision 8087
>> (http://code.pgadmin.org/trac/changeset/8087).
>>
>> I don't see an easy way to get around this. Either we drop that code, or
>> we explicitely state that pgAdmin works only with recent libpq (recent
>> meaning here 8.3+). Dave, did I misunderstand something? what's your
>> point of view on this issue?
>
> I don't see any reason why we should lose important functionality to
> ensure a modern pgAdmin will *build* with an old PostgreSQL. If we
> have to do that, it would become very difficult to know how any given
> build will behave.
>
>
