Training?

[Jerry Mahun]

I am new to SALSA and find it pretty overwhelming. Is there any training available for it or, if not, planned for the future?

[Clark Hughes]

Hi Jerry,

We have some future plans to get some introductory training videos online, but in the meantime we suggest new users take time to read through and exercise the examples in the user manual to develop familiarity and proficiency in SALSA. Give those a shot, and reach out with any specific questions.

Clark

[Jerry Mahun]

Tried following the Example 01 but it's definitely not a learn-it-on-your-own type of software. I've been struggling with SALSA for a few hours now and need to vent a bit.

SALSA is not very intuitive on a number of levels and that includes the documentation.

Rather than diving directly into a 3D GPS network, there need to be some simpler introductory projects. That would help to understand SALSA behavior and also make it easier to grasp how a project is built. I'm at a loss at how to even set up and execute a simple closed loop differential leveling circuit.

It's not clear what information I need in the first 50 pages or so in order to begin adjusting and analyzing networks. Does the difference between the Square Root Information Filter and  Conjugate Gradient Method make much difference to most users? Are all the a priori algorithms explanations necessary? Granted, it's all useful information ...  once you're more familiar with the software. Put it in an appendix. To slog through all of this on the way to learning how to set up a simple network makes SALSA seem  difficult to master (maybe it is?).

I got StarNet set up and adjusting networks relatively quickly. StarNet provides a number of different example projects from simple to complex which helped a lot. It also has, let's face it, a much friendlier and easier to navigate interface. Granted SALSA is free compared to StarNet, but with StarNat I was up and running in less than a day. and felt confident with it.

When software gets complex, there's a tendency to either shirk on the documentation or make it overly detailed from a developer's view point. What you need are two documents: (1)  How to Use SALSA with examples, and (2) SALSA Design, Components, Algorithms, Optimization, etc. The learning curve is just too steep.

Done venting.

[Clark Hughes]

Jerry,

Thank you for the thoughtful input. You're absolutely right in that we should reorganize the manual; the first example shouldn't start 50 pages in. A model that we've used in some of our other recent projects is to separate the algorithmic documentation (AKA "math spec") from the user documentation. This is essentially what you're hitting on in your last paragraph, and you're right; this model allows for light-weight documentation for users who may not desire a fundamental understanding of how the software works and/or how their solution is determined. This reorganization has been on our to-do list for a long while; thank you for nudging us on this...

SALSA is 3D, always. That is a fundamental design choice that we made a decade ago because the application was developed for (and funded by) geodetic survey applications. While SALSA will properly account for nuanced differences in geoid undulation and gravity deflections in adjusting your complex network of GNSS vectors, first-order leveling, and total station data, using SALSA to solve a trivial leveling loop is a real pain; it's not a 1D solver. (It can be done but requires adding constraints for the horizontal components of all points in the network.) Over the years, we've considered whether and how to support 1D and 2D networks; with some imagination it's probably doable without fundament changes under the hood.

Again, thank you for the thoughtful feedback.

Clark