Ignore specific message

[Wookey]

On Tue 28 Mar, John Halleck wrote:
> On Tue, 28 Mar 2000, Andy Waddington on Survey stuff wrote:
> 
> > Date: Tue, 28 Mar 2000 12:46:05 BST
> > From: Andy Waddington on Survey stuff <Survex@pennine.demon.co.uk>
> > To: survex@survex.com
> > Subject: re: Ignore specific message
> > 
> > Wookey discusses fixes to turn off "Fixed point not used" and/or "point
> > already fixed" errors.
> 
>   My personal opinion, though nobody asked, is that the first is not
>   intrinsicly an error, but a warning note that you should be able
>   to choose or not.

Agred, but should it be chosen as a 'warning level' a-la compilers
)specific to a run), or as a feature of the dataset (specific to
particular points in a dataset), or both/either?

>   The second seems to be serious...  and I think I'm missing why 
>   the data should have this problem.  (If someone could enlighten me
>   [possibly by private email instead of the list] I'd appreciate it.]

It happens if (for example) you have an overall file that fixes all the
entrance locations, but you also have the fixes in indivdual cave
overview files (because they want to be able to generate them on their
own with real-world co-ordinates, and don't like all the 'Fixed point not
used' warnings they get doing it the other way :-). 

Having them in two places causes a maintainance problem, but in fact,
stepping back even further, the only reason we do it this way is because
we need to separate surface an underground data for diplay purposes, but
we want to preserve the entrance location derived from the full dataset
even when only processing partial datasets. Once we have a mechanism for
toggling display of surface legs (new 3d format) then we can always just
run the whole dataset and this 'problem' will disappear. (the 'unused
point' one won't though.


>   In recent conversation with Olly Betts, he made statements about possibly
>   making the changes to the way fixed points were handled, since in this
>   day of GPS you might have several GPS fixed points, and they have uncertainty
>   that would have to be accounted for...

The *FIX syntax already allows for X,Y,Z SDs to be specified after the
co=ordinates, but this info is not yet used. Presumably you were
discussing the sums implementation. I'm really just thinking about the
data-format/user interface implementation.

Thinking about it I'm not sure that this is the right answer, as I don't
want to have to specify 3 extra numbers on every single GPS entry. I just
want to say they are GPS entries and get a ensible default, or perhaps
even specify a GPS entry and a time (SDs decrease with time of logging),
or some other quality indicator.

I suppose what this amounts to is DATA styles for points as well as
legs....maybe that's the way to go.

>   So... I suspect that the discussion might become a lot more interesting
>   when Olly gets back...

yes, but in the meantime we can establish what we think is correct
behaviour/what problems people have/relevant features they would like. So
long as said features don't include any sums then other people are
capable fixing the code accordingly (I avoid the bits with the sums in
like the plague!)


> > One workaround for the "fix not used" error is to add a zero-length
> > survey leg to each fixed point - messy, but then it depends whether
> > you are happier ignoring kludges in your data, or screeds of
> > ignorable warnings. Ignoring warnings in transient output is not a
> > problem, until there is a single error hidden somewhere in the great
> > screed of warnings that scroll past.
> 
>   In the standard (Matrix, Normal) equations, it is possible to deal with
>   multiple (weighted) fixed points rather trivally.   Since that is the
>   matrix that the Cholesky Factorization in servex is being done on, I don't
>   see why adding zero length survey legs would be usefull.  (Except for
>   some explainitory purposes.)

It's useful because it stops the error saying you didn't use the fixed
point, even though you used it in a manner which reduces to exactly what
you had before.

Wookey
-- 
Aleph One Ltd, Bottisham, CAMBRIDGE, CB5 9BA, UK  Tel (00 44) 1223 811679
work: http://www.aleph1.co.uk/     play: http://www.chaos.org.uk/~wookey/