Comments and Ratings

I found the source of my problem: My test files names did not begun with "Test"

Nick, thanks for the error report about assertElementsAlmostEqual and assertVectorsAlmostEqual. I uploaded a new version (2.0.1) today that fixes the problem.

If you are putting multiple tests in a single file, don't start any function names with "setup". The following code fails. Remove setup and it works.

function test_suite = testFliplr
initTestSuite;
end

function testFliplrMatrix
in = magic(3);
assertEqual(fliplr(in), in(:, [3 2 1]));
end

function testFliplrVector
assertEqual(fliplr([1 4 10]), [10 4 1]);
end

function test3
[a b] = setupdata;
assertEqual(a,1);
assertEqual(b,2);
end

function [a b] = setupdata
a = 1;
b = 2;
end

I think there is a divide by inf in computing the relative tolerance. Using the 'absolute' flag provides the correct result.

Great work! saved me some precious time...

Ryan, I decided not to do the sprintf-style conversion specifiers in the assert functions, as least for now. My experience designing syntaxes for MATLAB and toolbox functions makes me shy away from a syntax design with an indefinite set of trailing arguments, as in sprintf. Such a syntax design makes it difficult to expand the syntax to support other options in the future. For now, you can just inline a call to sprintf directly in the call to the assertion functions.

Ryan and Guido, the Help Browser team has identified the problem and is trying to fix the problem for the next MATLAB release. In the meantime, you can try this workaround: Put the MTEST directory after the MATLAB directories in your path, instead of before them.

Ryan, thanks for the enhancement suggestion. I have received several good suggestions, and I plan to work on an MTEST update soon.

This is very nicely written, and useful. For those MATLAB coders that desire to learn better practices.

This is extremely useful, but since it requires a function definition, you must either use lambdas, which don't allow procedural code (ie, a list of statements), define a local function, which means you have to turn your script into a function file (sometimes not desirable when dealing with workspace variables), or create a new file (lots of throwaway files with simple routines).

I wonder if it would be possible to add the following functionality (either in the same function or through a different one) :

while timeit %returns true when a new measurement is needed
  ... my procedural code ...
  ... multiple lines, will run as many times as timeit() sees fit ...
end
result = timeit; %next call: return result. reset state.

As you can see, this sort of usage would be tremendously useful to just profile a piece of code with no extra details -- just wrap it in a loop. Of course, it's only fair if you conclude it's not worth the additional effort :)

Steve, thank you for MTEST available I find it very useful.

Nevertheless I encountered a problem involving the mtest help-file. Choosing "MTEST Unit

Test Framework" in the help window causes Matlab to freeze. No input (mouse and keyboard) is

accepted and the application needs to be terminated by the task-manager. This problem can be

replicated by different users on different machines.

Details to replicate the problem:

- Copy Mtest to <mtest-dir>
- "Add Path with Subfolders" <mtest-dir>
- Open help in Matlab
- click on "MTEST Unit Test Framework"
=> Freeze

All other mtest help entries work as expected.

Matlab version: R2009a, 7.8.0.324 (win32)
OS: Windows XP Professional, 5.1.2600 SP3

It is really more a nuisance than a no-go but maybe you can track down the problem.

Thank you & kind regards
Guido

What can I say... Thank you for building and providing this. This should be part of the base MATLAB package!

Paolo, thanks for the suggestion. I'll look into it.

Great! helped a lot for me on learning connected components.

Excellent

After implementing Tarbotons Dinf algorithm for calculating flow directions I finally stumpled upon your toolbox, which is a lot faster. Thanks for the implementation, maybe if you add a keyword 'flow direction' it could save some other users some work ;-)

Take a look at 'colorGray' which has similar functionality but with "rainbow" like color spectrum.

its not qorking for me.. i use matlab 13 sp1
Warning: Input arguments must be scalar.
> In C:\Documents and Settings\501633799\Desktop\New Folder\patent_implementation\vislabels.m at line 27
Warning: Input arguments must be scalar.
> In C:\Documents and Settings\501633799\Desktop\New Folder\patent_implementation\vislabels.m at line 27
??? In an assignment A(I) = B, a matrix A cannot be resized.

Error in ==> C:\Documents and Settings\501633799\Desktop\New Folder\patent_implementation\vislabels.m
On line 28 ==> I(L == 0) = background_shade;

Great tool. Now I can paint NaN's in black on imagesc.

Thanks

Great tool work very well for what a needed to do.

Simple and nice one

Hi Steve,

areas of interest to geographers often are catchments which commonly have irregular shape. Beyond the watersheds there is often a lack of data. Hence, these values outside the catchment are NaNs. While NaNs inside the catchment may be interpolated the NaNs outside can merely be extrapolated.

Hence, I tried following. I replaced the NaNs in the DEM by -Inf. That works except that it produces an erroneous flow direction on the watershed pixels.

Best regards,
Wolfgang

Hi Steve,

great contribution! Users of older Matlab versions (R14) need to replace the function hypot(s1,s2) by
s = sqrt(abs(s1).^2 + abs(s2).^2)

One critique: The functions don't handle NaNs.

Best regards, Wolfgang

Simple, useful and well styled and coded. Did something in my existing code similar but this is a bit faster and less memory. Steve, you might want to note in the documentation that if you have a double input outside the range of 0-1 (soemthing you typically use imagesc to display) that you would want to run mat2gray on it before passing to this function. A useful further enhancement would be either to auto detect and apply this or have an optional switch.