From: Jerry <lan...@qw...> - 2008-07-21 00:46:34
|
On Jul 18, 2008, at 11:17 AM, Alan W. Irwin wrote: > On 2008-07-16 17:55-0600 Orion Poplawski wrote: > >> Alan W. Irwin wrote: >>> Orion, how does the Ada component of ctest work on Fedora (which >>> presumably >>> also has gnat-4.3 or later)? >> >> Testing latest svn with Fedora rawhide with gnat 4.3.1. Compiles >> fine but ada test x21a has now consumed more than 8 minutes of CPU >> time and appears hung. I'll check back in the morning... > > I guess not all gnat-4.3 packages are the same because I get > results that > are different from both Orion's results and Jerry's! > > It turns out I do have access to gnat-4.3 for Debian testing so I > installed > that, and x21a runs without problems for me contrary to Orion's > result but > in agreement with Jerry's result. However, when I look at the > second and > third pages, the Ada results for the first three of the 6 plots on > each page > are garbage. (Those plots should agree fairly closely with the > last 3 plots > on each of those pages.) > > Jerry, do you confirm those garbage plots for example 21 on your > platform? > If so, then my guess is when you fix this issue, Orion's issue will > also go > away. Note, the garbage results for Ada are a separate issue from the > reformatting of these Ada plots that should be done in any case to > conform > to the new C template results for example 21. Yes, I get garbage for the same plots as you--looks like some sort of indexing problem around the periphery of the data arrays. I apologize for submitting this in this state. I don't have any idea what might be Orion's problem with the long CPU times. > > The other difference I have with both Orion and Jerry for gnat-4.3 is > that example 19 (both thick and thin) gives me the following error. > > raised STORAGE_ERROR : stack overflow (or erroneous memory access) > > I also saw this same error for my previous gnat-4.1 platform. > > Andrew, do you see this issue on the various gnat platforms > accessible to > you or am I the only one seeing it? > > For those who sail through example 19 without this error, do your > Ada results > agree exactly with the corresponding C results? For me, x19a runs without reporting any errors and makes the same Postscript as x19c. Jerry > > Alan > > |