How to load, update and save data neatly

I have a variable saved to a .mat file which I wish to load in, update (extend) and save out again to the same file.

Obviously I can just call load with no output argument and then save. But people generally seem to say that

s = load( someFilename );

is far better than using the no output argument form. This I understand.

What I am finding difficult to do neatly is the saving back of my updated data. I can update easily as:

s.myVar = someFunction( s.myVar );

but then I can't just do:

save( someFilename, 's' )

as that will leave me with a structure in my file which would not then match the original format. I also can't do:

save( someFilename, 's.myVar' )

as this is clearly nonsense.

So do I really have to do the following?:

myVar = s.myVar;
save( someFilename, 'myVar' )

If so this seems to obviate the purpose of calling load with an output argument in the first place.

Answer 1

[Star Strider]

See if the matfile function does what you want. (I’m not certain when it was introduced, but it’s in R2014a.)

[Adam]

I'll give it a try. I have used matfile quite a few times, but it hadn't occured to me for this purpose. I generally use it when my data is too large to load into memory.

[Adam]

m = matfile( someFilename, 'Writable', true );
myVar = m.myVar;
myVar = update( myVar );
m.myVar = myVar;

does work and does at least relieve me of the need to have potentially unknown variables arriving in my workspace as I would with the no output form of 'save'.

Sadly, just updating the variable where it sits, in the matfile, does not work since it is of a custom class and I am resizing it.

[Star Strider]

Yes, according to the documentation, it doesn’t support user-defined classes. Sorry.

Answer 2

[Guillaume]

save(someFilename, '-struct', 's');

Will save the fields as individual variables (it's in the documentation of save!)

You can even specify which fields you want to save:

save(someFilename, '-struct', 's', 'fieldname1', 'fieldname2, ...);

[Adam]

Ah yes, so it is. I never expanded the 'variables' hyperlink as it didn't seem promising and I thought I knew what variables are! I also didn't notice the example lower down the page with the rather obvious titling :(