On Monday 30 November 2015 15.52:41 jan matejek wrote:
hello,
On 28.11.2015 21:14, Bruno Friedmann wrote:
I'm finishing the packaging of python3-csvkit.
figured it out. The problem is that the csvsql script insists on adding stdin if it's not a tty; that decision is questionable at best, and causes a failure in buildservice because stdin is redirected to some sort of empty input, which trips the script up. (i presume the same happens in koji)
Exactly, was my suspicion but didn't find a way to fix that. I'm still learning :-)
Indeed you can reproduce the same failure by adding "< /dev/null" to the csvsql command.
The attached patch simply reverses this decision and ignores stdin; note that it causes one of the tests fail because it is checking that stdin is in the list of inputs (apparently this decision is deliberate and checked for).
Ok thanks so much for the patch. I've improve it by also commented out the assert that was failing.
Ultimately it's up to you how to fix the problem; i presume you could ask upstream, or supply some other sort of input other than the redirected-stdin. Now we have good failing and a patch I will open that issue upstream and see what's their thought about it.
Also please note that the "random warning" and the subsequent backtrace is completely unrelated and you can safely ignore it. I'll try to do something about the backtrace at least; the tweak package is still experimental.
Nice to heard you will fix it, actually it does its jobs, but the output is kinda spam on every python build :-))
regards m.
Again thanks a lot, I'm no able to push the package further.
The spec (with test line commented is located here) https://build.opensuse.org/package/show/home:bruno_friedmann:branches:devel:...
-- Bruno Friedmann Ioda-Net Sàrl www.ioda-net.ch openSUSE Member & Board, fsfe fellowship GPG KEY : D5C9B751C4653227 irc: tigerfoot