Re: MADlib Debugging (elastic_net) - MADLIB-1068

2017-08-07 Thread Nandish Jayaram
Nice catch, thanks Ed! NJ On Mon, Aug 7, 2017 at 2:07 PM, Ed Espino wrote: > I have resolved MADLIB-1068 with a workaround (use GCC 4 and not the > default GCC 5) for Ubuntu 16.04. I have created MADLIB-1145 to track the > GCC 5 issue. > > Happy Monday, > -=e > > On Fri, Aug

Re: MADlib Debugging (elastic_net) - MADLIB-1068

2017-08-07 Thread Ed Espino
I have resolved MADLIB-1068 with a workaround (use GCC 4 and not the default GCC 5) for Ubuntu 16.04. I have created MADLIB-1145 to track the GCC 5 issue. Happy Monday, -=e On Fri, Aug 4, 2017 at 11:11 AM, Ed Espino wrote: > FYI: I have managed to get the issue to reproduce

Re: MADlib Debugging (elastic_net) - MADLIB-1068

2017-08-04 Thread Ed Espino
FYI: I have managed to get the issue to reproduce in gdb! It was relatively painless. I will be adding my debugging environment and notes to the dev list for future reference. I have come to understand Ubuntu isn't a supported platform. In the spirit of growing and fostering MADlib adoption, I'll

MADlib Debugging (elastic_net) - MADLIB-1068

2017-08-03 Thread Ed Espino
For MADLIB-1068, I have reproduced the elastic_net issue which causes a crash (core dump) in PostgreSQL 9.6 on Ubuntu 16.04.03 using MADlib master. Are there any general debugging techniques used to help track down these types of issues? I will be fiddling around with the elastic_net trying to