Re: [PATCHES] 8.0.0b1: configure.in cygipc

Attached is a cygipc patch against 8.0.0p1 and 7.4.5:

The cygipc patch should also be applied to the 7.4.x branch (if there
will be some more releases):

cygipc is deprecated in cygwin, replaced by cygserver
added test to check for running cygserver and the required
server in $CYGWIN. Otherwise build for cygipc, but print a warning.

replaced
AC_CHECK_LIB(cygipc, shmget)
by a better logic to detect cygserver, and only use cygipc if
cygserver fails.

******************************

BTW: My current set of CYGWIN-PATCHES I'm working on is at
http://xarch.tu-graz.ac.at/home/rurban/software/cygwin/postgresql/postgresql-7.4.5-1/
and
http://xarch.tu-graz.ac.at/home/rurban/software/cygwin/postgresql/postgresql-8.0.0beta1-1/
but the 8.0.0b1 patchset is not finished yet.
--
Reini Urban
http://xarch.tu-graz.ac.at/home/rurban/

Reini Urban wrote:
> cygipc is deprecated in cygwin, replaced by cygserver
> added test to check for running cygserver and the required
> server in $CYGWIN. Otherwise build for cygipc, but print a
> warning.

configure checks the build environment, not the run time environment.
It's completely irrelevant whether cygserver is running when building
PostgreSQL.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut schrieb:
> Reini Urban wrote:
>> cygipc is deprecated in cygwin, replaced by cygserver
>> added test to check for running cygserver and the required
>> server in $CYGWIN. Otherwise build for cygipc, but print a
>>warning.
>
> configure checks the build environment, not the run time environment.
> It's completely irrelevant whether cygserver is running when building
> PostgreSQL.

I know, but I didn't want to indroduce yet another --with-cygipc
configure switch, and the run-time detection is just decent a helper.
But my patch is still valid:
The current cygipc check is wrong when cygserver is running. If
cygserver is not running, it displays a warning when using the cygipc
library.

Later we really should check the cygwin release version and decide on
that. That will require `uname -r` parsing.
Ok?
--
Reini Urban
http://xarch.tu-graz.ac.at/home/rurban/

Reini Urban wrote:
> But my patch is still valid:
> The current cygipc check is wrong when cygserver is running. If
> cygserver is not running, it displays a warning when using the cygipc
> library.

What would be wrong with using cygserver when available (!= when
running), else cygipc?

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut schrieb:
> Reini Urban wrote:
>>But my patch is still valid:
>> The current cygipc check is wrong when cygserver is running. If
>>cygserver is not running, it displays a warning when using the cygipc
>>library.
>
> What would be wrong with using cygserver when available (!= when
> running), else cygipc?

I did that in my patch.
The problem is how to detect cygserver. I did it with checking against a
running cygserver. The other possibility would be to check against an
installed cygserver binary.

I favored the first.
Of course one should de-install cygipc, but I forgot that. Maybe others
will also.
It does not harm to build against the old cygipc daemon, but only if
cygserver is NOT running.
--
Reini Urban
http://xarch.tu-graz.ac.at/home/rurban/