Print

---

Paul,

I just looked at the source code for "reduce."  It seems there is a
global variable that gets returned from main(), ReturnCodeGlobal,
which is set to 0 by default.  However, further down in the code is
this (in the "processPDBfile" function):

// adjust cliques

	 std::list< std::list<MoverPtr> > cc_list = clst.cliques();
	//cerr << "start: " << cc_list.size() << endl;
	 for (std::list< std::list<MoverPtr> >::iterator cc =
cc_list.begin(); cc != cc_list.end(); ++cc) {
		//cerr << "start2" << endl;
	    int nscnt = xyz.orientClique(*cc, ExhaustiveLimit);
		if (nscnt > 0) { Tally._num_adj += nscnt; }
	    else { // too many permutations, make note
	      ReturnCodeGlobal = ABANDONED_RC;
	    }
	 }

Where ABANDONED_RC is elsewhere defined as 1.

This is the only obvious place in the "reduce" code I can find where
ReturnCodeGlobal is set to something other than 0.  I'm not sure what
they mean (in the comment) by "too many permutations," though.  Do you
think this is a fatal error that calls for not running "probe"?
Otherwise, I might try to just work around this by modifying the Coot
source to ignore the return value of "reduce."

Thanks again,
Tom

On Mon, Oct 12, 2009 at 1:54 PM, Paul Emsley <[log in to unmask]> wrote:
> Thomas Cleveland wrote:
>>
>> Bernhard,
>>
>> I have tried this using several versions of coot.  I tried it first in
>> the coot that came packaged with ccp4 (Coot 0.5.2).  I then tried
>> WinCoot-0.6-pre-1-revision-2411, which is one of the later revisions
>> of 0.6-pre, just to see if a more recent version would work.  As far
>> as I can tell, I am using the correct versions of probe and reduce.
>>
>> I am attaching a file with the relevant portion of the output
>> (starting with when I run "probe clashes," up until it fails).
>>
>
> OK, so that's some meat to chew on...
> Interesting things AFAICS:
>
> BL WARNING:: reduce didnt run ok, so stop here!
>
>
> It is convention in the Unix world that a program that runs without errors
> exits with a status of 0.  This is not happening when you run reduce from
> within WinCoot - it is not clear to me why that is. I raise a doubt over
> your installation of reduce.  Or it might be that reduce exits with a random
> status in Windows and you are unlucky.
>
> Regards,
>
> Paul.
>
>