On Mon, Mar 9, 2009 at 7:51 PM, Martin Budaj <m.budaj@gmail.com> wrote:
I think that it is not related to the bug in the dataset of Martin Sluka. Could you send (to list or to me) some minimal sample?
It may be difficult to reduce the problem to a minimal sample; if I change anything the survey compiles without errors. I have tried: Changing the scale (problem only at 1:1000) Commenting out scraps (problem goes away if any scrap is commented out) Using different symbol set (problem goes away if default symbols are used rather than my customised ones) Commenting out custom symbol definitions (problem goes away if any custom symbol definition is commented out) Selecting more maps (a map that contains the 'problem' map compiles at 1:1000 without errors) Removing non-selected maps (if comment out in the master .th file for the area maps that aren't selected in the .thconfig file that causes the problem, it compiles without errors). Compiling other parts of the system or area works OK. So - it looks like if you want to see the problem for yourself, I'll have to send you my whole data set (which I'm quite happy to do) along with my symbol definitions.
The new snapshot has changed (improved) the MetaPost and TeX macros, so there should be a problem.
Unfortunately, latest fixes are not yet released.
FYI I'm using 5.2.7 on WinXP Duncan.