[prev in list] [next in list] [prev in thread] [next in thread] 

List:       graphviz-interest
Subject:    Re: [graphviz-interest] ccomps segfault on requesting too many
From:       "Emden R. Gansner" <erg () research ! att ! com>
Date:       2011-12-14 17:48:26
Message-ID: 4EE8E16A.9050209 () research ! att ! com
[Download RAW message or body]

On 12/14/11 12:11 PM, Joseph Bylund wrote:
> 2) all the unconnected components, which I request as:
>     ccomps -zX#1-1000 program.map > program_unconnected.map
> I use 1000, because I want to make sure I catch them all.  This 
> actually works as expected, but segfaults. If I send it -zX#1-114 
> (which is the actual number of components) it works as expected, but I 
> won't always know what this number is.
>
This was a silly error where ccomps trusted you to give a reasonable 
upper bound. It is now fixed, with the changes appearing in
tomorrow's packages. By the way, if you don't know the upper bound, you 
can always use -zX#1- which serve as a workaround.
> A second related question.  There are some subroutines which make 
> calls to things in the main component, but are never called by things 
> in the main component.  I am also interested in identifying these, is 
> there some method similar to ccomps which could tell me this information? 
ccomps assumes your graph is undirected. If there is any edge from a 
subroutine into the main component, that subroutine will be
part to the main component. If you are not currently getting this, you 
must have constructed your call graph to only use forward
edges.

     Emden

_______________________________________________
graphviz-interest@research.att.com
https://mailman.research.att.com/mailman/listinfo/graphviz-interest
[prev in list] [next in list] [prev in thread] [next in thread]