questions on *date and *export

[Olly Betts]

On Sat, Aug 26, 2006 at 10:18:31AM +0200, Thomas Holder wrote:
> Thanks for your replys, Wookey and Mark. Then *export is just for
> keeping a proper style.

I wouldn't say it's just stylistic.  Experience shows that mistying
surveys together is one of the more common errors, especially in large
surveying projects involving many people.

The idea behind *export is to help address this problem by providing a
way to document which stations are refindable - partly for use by
humans, but also in a way allows Survex to automatically check that
surveys are only connected at refindable stations.

Incidentally, if you don't use *export AT ALL, then you don't need
"*infer exports on".  I added "*infer exports on" so you can join
together two datasets when one uses *export and the other doesn't.

> Do you also know if multiple *date commands in the same *begin/*end
> block (and not only in the beginning but in between the data) will be
> valid with future survex versions, although the manuals says something else?

I've taken a look at the code.  Currently there's no policing of
any commands which are documented as needing to be at the start of
a survey, apart from *export (because there's also a technical reason
why we want *export to appear before any survey data).

I don't recall the exact reasoning behind which commands are
documented this way, so I'm not sure I can give a definitive
answer right now - I'd need to look at the mailing list archives
and the source code history.  I suspect it was just because we
felt it was "good style" to put the meta information together
at the start of a survey, so wanted to encourage people to do
it.

But I think you can rely on it being a warning at worst since
there's no doubt plenty of existing data out there which doesn't
comply with these restrictions!  Your example with wanting to
specify *date multiple times is reasonable enough.  Dave's suggestion
satisfies the documented requirement, but I'm not sure forcing people to
wrap such uses in *begin/*end actually benefits anybody.

Cheers,
    Olly