Deg invalid in units

[Olly Betts]

On Wed, Jan 11, 2017 at 11:01:45PM +0000, Footleg wrote:
> Thanks for that. Yes I agree updating the manual to match the behaviour
> makes most sense. I'll update my parser to match Survex actual behaviour
> too. I'm happy to provide all my test files to the Survex test suite.
> They'll be included in my source distribution for cave converter under a
> GPL license when I've finished writing them. I'll send you a link when my
> release goes out.

Technically I can't just take files from cave convertor and add them to
Survex, as cave convertor is GPLv3+ while Survex is GPLv2+ - I'd have to change
Survex to be GPLv3+ to use the two together.  But if you're happy to offer them
licensed under GPLv2 as well as v3 and later that's great.

> Since I added handling of file includes to my parser I've been testing it
> against the entire 3 counties system dataset, which has raised some
> interesting questions about what should be valid or maybe at least raise a
> warning. I added output of splays using anonymous to stations, which
> enables my converter to read files with splays flagged using flags in
> Survex, and convert them to splays using - instead of a named to station.
> This broke some files in the Leck Fell dataset where survey data was joined
> the both ends of a splay leg. Survex processed these without warnings,
> which I thought was a case where a warning could be useful as it seemed an
> unusual thing to do with a leg flagged as a splay. After I converted those
> named to stations to - characters it broke the connectivity of these data
> sets.

It seems reasonable to define splays as being a single leg with no
continuation, though I guess you could have multi-leg measurements to walls and
think of the whole thing as a splay.  That case could instead be flagged as a
series of "DUPLICATE" legs with only the final leg being "SPLAY", though that
is more cumbersome to type in.  So perhaps the definition should be one or more
legs flagged "SPLAY" with the final station not connecting to anything else and
not being a fixed point.  I'd lean towards the simpler definition unless we
get a report that this multi-leg case is actually common practice somewhere.

The end of a splay probably shouldn't be exported or marked as an entrance
either.

I don't have a copy of the three counties dataset to hand - what do these cases
look like exactly?  It sounds like you're saying these splay legs lead to more
non-splay survey data, which definitely seems bogus.

I think in a case like this where it's been accepted and handled in an fairly
obvious and reasonable way for a long time, it's better to warn about it than
reject the data.  Perhaps the "SPLAY" flag should be unset too.

Cheers,
    Olly