Here is an issue that took me too many days to unravel.

The order of source statements in a thconfig affects the outcome of the compilation.

I think the problem, or constraint, can be described as:

• Multi-line source statements must follow all single line source statements, else not all surveys will not be read correctly, or
• Single-line source statements that follow a multi-line source statement will not be read.

For example, a thconfig contains:

source ./Greenlinkcave/INDEXGreenlink.th

source ./swissmaidcave/INDEXSwissMaid.th

source

    #join the two cave surveys without an index file

    equate 133@jp.SwissM  I68@s.Greenlink

    equate 156@second.SwissM S45@ss.Greenlink

    equate 144@jp.SwissM H32@h.Greenlink       

endsource

source ./middleearthcave/INDEXMiddleEarth.th

source

    #join the two cave surveys without an index file

    equate F76@f.MiddleEarth F102@f2.Greenlink           

endsource

This results in a fatal error:

reading source files ... done

preprocessing database ... done

C:\Program Files\Therion\therion.exe: error -- ./middleearthcave/INDEXMiddleEarth.th [24] -- survey does not exist -- 74

Which confused me because each of these would compile alone an in combination via other thconfig files.

If I changed the order of the source statements, the ‘non-existent’ survey would change.

I tried 6.2.0 + fd5c1b6 and 6.17 and 6.13 all with similar effect (although I’m not that confident because the latter two being 32-bit and toggling between the 64-bit 6.2.0 variants results in duplicate therion programs on my Windows machine, thus I’m not sure I have ‘clean’ installs).

Too late, I tried putting all the single line sources first, followed by the multi-line sources, and viola, success.

So this seems like a bug, in that I understand Therion should read all sources before processing the survey and map input.  With sources, this seems not to be happening, and Therion is crashing inappropriately if the sources are not listed in the ‘correct’ order.

Bruce

PS: Yes, I know the example above is a rather messy way of treating sources.  I would not usually do this – it is a hack to work around an unfortunate decision made early in the project history.