agree23 wrote:Don't know if this should be in regression test or in technology...
Do you eValid guys have any thing that looks like "fuzz testing"...
I mean, some way to apply random inputs over and over and over in hopes of revealing a defect?
Yes, eValid's eV.Generate component can be used to do this.
With eV.Generate you specify a script template file that has parameters in it that will be filled in by eV.Generate before the script is dropped into the eValid playback engine. The parameters can specify anything in the script.
The values that are put into the template file that eV.Generate uses to create the playable script file are taken from a "value table" that enumerates, for each possible variable all of the possible values that variable can take on.
For example, you might have: "letter a b c d e f g" as one line in the value able and in that case the instances of "letter" in the template file can be given playback time values of a, b, c...
You can have any number of different defined parameters and for any of them any number of possible string values that could be assigned to them.
Once the template is built and the values table is populated you then invoke eV.generate in "random" mode directly from your eValid script. At playback time the values for the parameters in the template are selected at random using a uniform distribution from the lists of possible values given in the value table.
Run this over and over and you have very effective "fuzz test".
The eValid Support Team