Re: Compilations failing - Mailing list pgsql-interfaces

From Jeroen T. Vermeulen
Subject Re: Compilations failing
Date
Msg-id 6643.203.121.164.162.1159771898.squirrel@webmail.xs4all.nl
Whole thread Raw
In response to Re: Compilations failing  ("Andreas Gidlund" <vaxis@hotmail.com>)
List pgsql-interfaces
On Mon, October 2, 2006 02:36, Andreas Gidlund wrote:

> it will stop when
> compiling \src\backend\utils\mb\wchar.c, because there are two functions
> which cannot be compiled, pg_euc_mblen and pg_euc_dsplen. The reason is
> that
> they are declared both static and inlined, which is a syntax nmake
> appearantly does not support.

That's not nmake; that's the compiler.  The combination of "inline static"
sounds perfectly valid to me though--don't see why that would be a
problem.  Perhaps it's a simple matter of "static inline" vs. "inline
static?"


> Though, here comes a problem. When I tried to execute a query and hence
> recieve the result into a PGresult, the compiler complains about that the
> variable uses an undefined struct, pg_result.

That's odd...  Are you compiling postgres as "managed code," by any
chance?  That would explain runtime errors of this kind (which in C and
C++ should give you compile-time errors or just work), and it's also known
to cause lots of confusion and link errors with libpqxx.

The problem with "managed code" is that it's not really compatible with
C++, but this fact is apparently not made very obvious.  From what I hear,
the code can look similar but it's not possible to link "managed C++" to
C++ DLLs.  The situation with C may be similar.


Jeroen




pgsql-interfaces by date:

Previous
From: "Andreas Gidlund"
Date:
Subject: Re: Compilations failing
Next
From: "Andreas Gidlund"
Date:
Subject: Retrieve all the returned rows in one call