Thread: Support Python 3 tests under MSVC
Here are a couple of patches, one for master and one for the back branches from 9.5 to 10 to allow testing of plpython3 and associated contrib modules when building with MSVC. I've tested this out on a Windows 2016 machine with Visual Studio 2017. I'd like to apply these - they only affect the testing script vcregress.pl, so they should be very low risk. cheers andrew -- Andrew Dunstan https://www.2ndQuadrant.com PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Attachment
On 4/30/18 15:52, Andrew Dunstan wrote: > I'd like to apply these - they only affect the testing script > vcregress.pl, so they should be very low risk. In case there are concerns about maintaining a second copy of the "mangle" script: The obvious answer is to rewrite that script in Python. I tried that once upon a time but it got too complicated. -- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Interesting, this working fine more than year with CMake build version.
On Tue, 1 May 2018, 04:53 Andrew Dunstan, <andrew.dunstan@2ndquadrant.com> wrote:
Here are a couple of patches, one for master and one for the back
branches from 9.5 to 10 to allow testing of plpython3 and associated
contrib modules when building with MSVC.
I've tested this out on a Windows 2016 machine with Visual Studio 2017.
I'd like to apply these - they only affect the testing script
vcregress.pl, so they should be very low risk.
cheers
andrew
--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On Mon, Apr 30, 2018 at 4:52 PM, Peter Eisentraut <peter.eisentraut@2ndquadrant.com> wrote: > On 4/30/18 15:52, Andrew Dunstan wrote: >> I'd like to apply these - they only affect the testing script >> vcregress.pl, so they should be very low risk. > > In case there are concerns about maintaining a second copy of the > "mangle" script: The obvious answer is to rewrite that script in Python. > I tried that once upon a time but it got too complicated. > Well, as you can see I have rewritten it in perl :-) We could extract that into a perl package and centralize it, but I'm not sure it's worth the trouble. Here's a new version of the master patch. I wasn't able to resist the temptation to do a little cleaning up. In particular, we really don't need to call python to tell us which version we have - we can do what we do elsewhere and look to see what python build directory we have. I also removed the completely redundant first parameter of subdircheck(). cheers andrew -- Andrew Dunstan https://www.2ndQuadrant.com PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services