[pgrouting-dev] Bidirectional Integration bug

[Stephen Woodbridge]

Hi Razequl,

I did a little more research on this problem and added try-catch block 
in your C++ wrapper function which is reporting: std::bad_alloc

In general, I think for robustness we need all C++ code to include try 
blocks and clean up on exceptions to prevent server crashes and memory 
leaks on exceptions.

I will check in my code, so that will give you a head start. And I 
except that a simple code review and addition of try blocks at 
appropriate places will identify the issues.

Thanks,
   -Steve

On 5/7/2013 1:35 AM, Razequl Islam wrote:
> Hi Steve,
> I also want to have a look at the possible problem. I am trying to set
> up the environment on my machine. I have the previous version. What do I
> need to do to get 2.0 working? Should I set up the environment from the
> scratch or should an update will do. Please let me know how I should
> proceed. I can also do the debugging once I have the environment in place.
>
> Thanks.
> - Razequl
>
>
>
> On Mon, May 6, 2013 at 11:32 PM, Stephen Woodbridge
> <woodbri at swoodbridge.com <mailto:woodbri at swoodbridge.com>> wrote:
>
>     Hi Razequl,
>
>     I have been trying to integrate your bidirectional shortest path
>     code. I have merged the code into my pgrouting branch
>     "sew-devel-2_0", but it is crashing the back-end with a segv.
>
>     If you have time and want to look into this that would be great as I
>     would like to get this code working in 2.0.
>
>     I'm not sure if I found a bug in your code or I broke something when
>     I merged it, or for that matter that I even merged the correct code
>     :) because you sent email indicating that you fixed stuff. I merged
>     the code from master at https://github.com/zibon/__pgrouting
>     <https://github.com/zibon/pgrouting>
>
>     I also ran into an issue that when I merged you code into the code
>     library, that the trsp broke, then take you code out and it works
>     again. I assume that there is some kind of function name or global
>     initialization issue. This does not seem to be an issue for trsp now
>     that the bi-dir code is in a separate library.
>
>     This might give you a chance to get your development/test/debug
>     environment setup and to familiarize yourself with the new source
>     code layout.
>
>     Let me know if you have time and I'll help point the way. If you are
>     tied up with class work, etc that clearly takes priority.
>
>     Thanks,
>        -Steve
>
>