Survex 1.1.6 development snapshot uploaded

[David Loeffler]

On Wed, 2005-10-12 at 03:02 +0100, Olly Betts wrote:
> On Tue, Oct 11, 2005 at 03:01:23PM -0400, Philip Balister wrote:
> > I the gradient box I have dates scattered around 1933 to 1942 or so.
> 
> OK, I've tracked this down (thanks to an example from Philip).
> 
> The issue only affects 32 bit platforms and only for dates after
> 2004-01-10.
> 
> The *date command allows a range - usually both ends are the same.
> We take the average of the range for display and for dates after
> 2004-01-10 this average overflows and goes negative.  Then dividing
> by 2 gives a negative answer, which is before 1970!
> 
> I'll fix it in 1.1.7 - it looks like this will be quite soon to mop
> up a few problems from the large number of changes in 1.1.6, so
> please report more problems promptly!

OK, here's another bug for you. It seems that if the date information in
a survey is very fragmentary, then it won't offer the option of
colouring by date (this happens for all.svx in the CUCC Loser dataset).
This is fine in itself, although somewhat inconsistent; but here's the
bug. If you have a survey loaded in Aven with "colour by date" selected,
and you attempt to load up a .svx file directly (using the new
process-on-loading facility, which incidentally is a Very Good Thing
imho), and the new survey you're loading has no date information, you
get an immense flood of "Bug in program detected!" error dialogs (I had
150 of them on my screen at once) and as soon as you click OK in any of
them, Aven quits. Presumably this is because it's still in "colour by
date" mode and doesn't turn that off before trying to load the new
survey.

The error messages reported at the console look like this:

OpenGL error: invalid operation: call glNewList in function CreateList
aven: gfxcore.cc:2185: void GfxCore::SetColourFromDate(long int,
double): Assertion `how_far >= 0.0' failed.
new list #5: 389... Abnormal termination

David