<div class="gmail_quote">On Sat, Aug 13, 2011 at 17:30, Barry Smith <span dir="ltr"><<a href="mailto:bsmith@mcs.anl.gov">bsmith@mcs.anl.gov</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Yes we have to fix our testing process so one doesn't need to constantly change the output files.</blockquote></div><br><div>Would it be hopeless to have the view routines always write JSON and then format it using a filter that doesn't have detailed semantic knowledge of each part? Alternatively, a lot of the output could be parsed in current form without _that_ much semantic knowledge (petscplot needed this), so we could recover numeric values and do some sort of fuzzy comparisons when deciding how severe a mismatch is. The comparison for convergence monitors needs semantic knowledge that it is a convergence monitor anyway.</div>