On Friday 03 June 2005 01:54, Dave Page wrote:
> As mentioned in another thread yesterday, I have tweaked this patch -
> the new version is attached. To apply, apply the patch as normal, and
> then rename all the src/ducttape/test_*_* scripts to
> src/ducttape/test_*_*.in.
>
> As I also mentioned yesterday, I don't expect this to be applied now as
> I realise a release is in progress - I'm just putting it 'out there' for
> anyone that is interested and any comments/objections/flames etc :-)


Thanks for all of your work, and you and your teams eagerness to get win32 
support working.  I have a new testing framework I'm working on implementing 
that should make the ducttape tests act more like the pg_regression tests.  
This new test framework should allow us to create easily customizable and 
distributed testing environments.  Once complete I'd like to see slony have a 
system simular (or perhaps connected too) the pgbuildfarm.org to facilitate 
cross version/arch compatibility testing, since as we all know currently it's 
up to what ever resources us developers have at hand to test on.



>
> Regards, Dave.
>
> > -----Original Message-----
> > From: slony1-general-bounces at gborg.postgresql.org
> > [mailto:slony1-general-bounces at gborg.postgresql.org] On
> > Behalf Of Dave Page
> > Sent: 02 June 2005 16:19
> > To: slony1-general at gborg.postgresql.org
> > Subject: [Slony1-general] [PATCH] Win32/Msys Ducttape tests
> >
> > Hi,
> >
> > The attached patch rewrites the ducttape tests during the
> > build to work
> > correctly on Msys/MinGW on Win32. I am not a make expert by
> > any stretch
> > of the imagination, so if there's a more appropriate way of
> > doing this,
> > please let me know.
> >
> > The following changes are made, provided $(host_os) == mingw32:
> >
> > - xterm is replaced with /bin/rxvt as there is no xterm in Msys by
> > default.
> >
> > - /bin/which is specified. Mingw has it's own 'which' in /mingw/bin,
> > however in my experience this often crashes. The Msys one is a simple
> > script that calls 'type -p $1' and works reliably. The order
> > of these 2
> > directories in the path is affected by the value of $MSYSTEM, so it
> > seems easiest just to hard code the correct path.
> >
> > Also, the following change seems reasonable for all platforms.
> >
> > - Use psql's -o option to output data dumps to files in
> > compare_pgbench_dumps. This avoids using the pager which is
> > likely to be
> > C:\windows\System32\more.com on Win32 and cannot handle more
> > than 8MB of
> > data at once. Even the MinGW/Msys pager (less.exe) fails to work
> > properly, outputting spurious extra spaces thus requiring diff -b
> > instead of plain diff.
> >
> > Regards, Dave.

-- 
Darcy Buskermolen
Wavefire Technologies Corp.

http://www.wavefire.com
ph: 250.717.0200
fx: 250.763.1759