[therion] Re: Survex vs Therion

Stacho Mudrak s.m at speleo.sk
Wed Apr 28 10:50:19 CEST 2004


> I did actually try to start to encourage a bit of cooperation a while
> ago - in particular regarding the details of therion's "charset" command
> which Survex could usefully implement.  But I didn't get a useful
> response...


I'm sorry. Are there any problems concernign the character sets?

> But if you just want to hide cavern inside therion, it wouldn't be
> hard to compile in messages.  You could even have your own set of
> "translations" providing easy to machine parse output for E-W range,
> etc.  The pieces needed to provide a mechanism to compile in a single
> translation are already mostly there.


I've tried this - but without success. The problem were not the messages, but entire compilation. It's very dependent on a compiler configuration - and I'm not very familiar with automake and autoconf.

> Nobody's mentioned to me that they're waiting for loop closure to
> become a library.  I've got a list of things to do, and I have to decide
> on a sensible order.  So far, making loop closure a library has been
> well down the list because there are other items which will do more to
> improve the user's "survex experience".


You're right. But therion worked as it worked for a long time, and we have no problems with it, until we've urgently needed a simple standalone windows installation.

> If the code would be useful, perhaps this should be higher priority...


It would helped us a lot.

> A consumer grade GPS probably won't report them (some report DOP I
> think).  Survey grade units will though, and since the loop closure
> algorithm can make use of the information, it's a shame not to be
> able to specify it when you have it.


OK, I see you're right. I'll add these covariances back. I just thought, nobody has never used them... Or do you really know somebody?

> Sadly the idea of loop closure statistics for "complete loops" seems
> obvious but isn't really very meaningful.  The closure is done on those
> segments, and to report information for "complete loops", you have to
> combine those segments in some arbitrary way.  And by combining them
> you lose information - in particular large errors which indicate
> possible blunders will be less obvious.
>
> "Minimal number of shots" seems particularly arbitrary.  "Minimum
> length" seems a bit more justifiable.  But either is going to prefer to
> pick out the survey loop around the walls of the chamber near the
> entrance rather than your magnificent new closure involving several kms
> of survey.
>
> The most sensible approach is to display the information graphically.
> Then where the "complete loops" are is no longer an issue.  And if you
> want to know the closure on a particular loop, you can highlight it with
> the mouse and ask.


This is interesting. I've never thought about it this way...  I'll probably try some examples to see, whether it really workes this way... I'm very curious.

Thanks a lot for your explanation and comments.

S.





More information about the Therion mailing list