> 2) Avalanche is good in discovering errors on 'sad paths' - i. e. when > the program is run on some kind of malformed input. And one can usually > stick quite enough malformation even into short files:)
That's going to explore the problems in input parsing fairly well, but layers of processing that lie behind that are another matter. For the stuff I work on, a very large percentage of all the possible input byte sequences of any particular length will be invalid input, and a large fraction of those will be recognised as invalid by the input parser. As far as I understand your preprint, Avalanche seems more likely to discover the input sequences that result in errors after very little processing than those that produce errors after more processing time? best, -- John Dallman Parasolid Porting Engineer Siemens Product Lifecycle Management Software Industry Sector 46 Regent Street, Cambridge, CB2 1DP United Kingdom Tel: +44-1223-371554 [email protected] www.siemens.com/plm > -----Original Message----- > From: Ildar Isaev [mailto:[email protected]] > Sent: Wednesday, May 26, 2010 4:08 PM > To: Dallman, John > Cc: [email protected] > Subject: Re: [Valgrind-users] Valgrind tool for generating 'inputs of > death' > > 1) 10 Kb is not that terribly big either - it still makes sense trying > to run the analysis and examine the results > > 2) Avalanche is good in discovering errors on 'sad paths' - i. e. when > the program is run on some kind of malformed input. And one can usually > stick quite enough malformation even into short files:) > > 3) Don't be very strict to Avalanche - it is just a research, which > results greatly exceeded my expectations. So now I'm just thinking about > better ways of sharing my work and getting some feedback. Of course, > there's still a room for improvement. > > > This is fine with some kinds of data. One can make a smaller bitmap, > > or a shorter sound clip. But with what I do - accurate 3D shape > > representation - one can't get anything meaningful into 1KB or so. > > I just took a look at our directory of synthetic test parts, and > > there are some under 1KB, but they are mostly null cases. You > > start to get non-trivial data at 10KB or so sizes. This doesn't > > make Avalanche useless, but it does limit it fairly seriously. > > > > best, > > > > -- > > John Dallman > > Parasolid Porting Engineer > > ------------------------------------------------------------------------------ _______________________________________________ Valgrind-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/valgrind-users
