Survex 1.0.19 released

[Olly Betts]

Here we go again:

http://www.survex.com/download.html

No versions for RISC OS or MSDOS on 286 or earlier machines yet - I need
to get packing for a caving trip this weekend.  Hopefully I'll get them
updated next week.

The major change is the addition of support for reading raw survey data
in Compass format.  If you have Compass data, please try this out and
let me know how it goes.  See below for a few caveats.

Here's the usual summary of changes in 1.0.19:

* cavern: you can now process Compass format survey data (.DAT and .MAK files) -
  these can even be linked into a Survex dataset with "*include fulford.dat" or
  "*include lech.mak" to allow processing of mixed datasets.

* cavern: major speedup (up to 60%) when reading in survey data containing
  a lot of stations in the same survey hierarchy level (which is how Compass
  files are handled).

* cavern: don't warn about a compass reading on a plumbed leg if the
  plumb was inferred and the compass reading is zero.

* cavern: reduced memory usage in code which reads a survey station name.

* cavern: reported CPU time would go negative if cavern ran for more than
  about 36 minutes (highly unlikely but not impossible on a slow machine
  with a large, highly interconnect survey) - fixed.

* cavern: avoid unnecessary work when --percentage isn't specified.

* cavern: Writing "*:" at the start of a .svx file would crash cavern - fixed
  and added test case.  This is an error anyway, but it shouldn't cause a
  crash.

* Don't distribute unused toolbar icons.

* Documentation: minor updates.

A quick note on Compass support - this isn't perfect yet.  The currently
known shortcomings:

* No support for Compass leg flags - in particular legs flagged to be
  ignored completely with the X flag will be included which may well
  cause processing to fail...  With care you can remove them with
  a text editor (delete any line matching "#|X").

* Redundant back-compass and back-clino readings are currently ignored
  - fairly harmless in general.

* Link stations in Compass MAK files are ignored - if you're using them
  to save space/speed things up a little, the survey can still be
  processed successfully.  But link stations can also be used to allow
  processing of two surveys with clashing names - if the data relies on
  this, like-named stations will be connected together with messy
  results.

The first two will be fixed shortly.

The last is a trickier matter.  To fix it exactly would require much
deeper changes than I'm happy doing in the stable 1.0 series (the
changes to read Compass data are mostly self-contained, so any
bugs introduced are almost certain to only affect the reading of
Compass files).

Also there's an easy work-around - you can construct a top level .svx
file to replace the MAK file.  Even if you're regularly updating the
Compass data, the MAK files is likely to change infrequently and is
fairly short, so keeping the parallel .svx version updated shouldn't
be much of a burden.

Cheers,
Olly