Am I creating dynamic variable names?

Recently I came across a popular thread in MATLAB Answers: https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval#answer_236124

After reading this thread, I am still not sure I have created "dynamic variable names", which should be avoided as suggested in that thread.

I have a code snippet as following:

data = [1, 2, 3];
centerX = data(1);
centerY = data(2);
centerZ = data(3);
% And use these variables afterwards

Should I avoid creating those (probably dynamic) variable names, i.e., centerX, centerY, centerZ to use those data and simply/directly use data(1), data(2), data(3) instead?

[Stephen23]

"After reading this thread, I am still not sure I have created "dynamic variable names""

No, you have not.

Absolutely nothing in your code dynamically names any variables or accesses any variable names dynamically.

"Should I avoid creating those (probably dynamic) variable names,"

What makes you think that they are "probably" dynamically named, given that they are hardcoded variable names?

[hmhuang]

@Stephen Because of the examples given in https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval#answer_236124:

• matrix1, matrix2, matrix3, matrix4, ...
• test_20kmh, test_50kmh, test_80kmh, ...
• nameA, nameB, nameC, nameD, ...

[hmhuang]

@Stephen But now I understand those example given in that thread were dynamically created using eval, which is different from my hardcoded ones.

Answer 1

[Sulaymon Eshkabilov]

What you are doing here is not dynamic variable naming. You're assigning a new variable name from your already assigned data.

This example shows the dynamic variable naming and assigning values to the dynamically named variables (U and V):

for jj=1:5
    eval(['V' num2str(jj) '= jj'])
    eval(['U' num2str(jj) '= ' 'V' num2str(jj) '*jj'])
end

Which is NOT recommended to employ.

That is equivalent to:

V = 1:5;
U = V.*V;  

This one is the recommended one.

[hmhuang]

@Sulaymon Eshkabilov Thanks! One question: Why one would decide to use the expression with eval if they can use the much simpler and efficient syntax (V = 1:5; U = V.*V). The syntax with eval is complex for me in the first place...

[Walter Roberson]

It is common for people to load multiple files and to want to have a full variable named the same thing as the file.

[Sulaymon Eshkabilov]

Walter pinpointed one of the key reasons why the dynamic variable naming is employed. In addition to that, some users want to control/name variables in a specific (or sequential) order and a dynamic pace.

[Walter Roberson]

"There are 2 hard problems in computer science: cache invalidation, naming things, and off-by-1 errors."

[hmhuang]

"It is common for people to load multiple files and to want to have a full variable named the same thing as the file."

@Walter Roberson Can you please give an simple example?

[Walter Roberson]

dinfo = dir('*.png');
filenames = fullfile({dinfo.folder}, {dinfo.name});
for K = 1 : length(filenames)
   thisfile = filenames{K};
   thisimage = imread(thisfile);
   [~, basename, ~] = fileparts(thisfile);   %just the name with no folder or extension

At this point, the user wants to use the content of basename (so, just the name part of the image) as the name of the variable to write to. For example if the files were 001IMG.png 002IMG.png 003IMG.png then afterwards they would want variables named 001IMG, 002IMG, and 003IMG .

(No, those are not capital-O at the beginning of the file names, those are zeros... but people would want the variables named starting with a number anyhow.)

[Walter Roberson]

Another use that shows up from time to time, is if people have a file that contains the names of variables and associated values. For example they might have an .xlsx file that contains

pressure_ammonia, 0.73
mols_ammonia, 548.19
ambient temperature, 280.4

and they would want out of this

pressure_ammonia = 0.73
mols_ammonia = 548.19
ambient temperature = 280.4

You may notice that ambient temperature contains a space in it. That is irrelevant to the people who want to do this: they would want a variable name with a space in the middle of it to be created.

[hmhuang]

@Walter Roberson Thanks for your detailed answer! One more question: so it seems that in this use case we can even "have a full variable named the same thing as the file" without using eval, as your example shown, right?

dinfo = dir('*.png');
filenames = fullfile({dinfo.folder}, {dinfo.name});
for K = 1 : length(filenames)
   thisfile = filenames{K};
   thisimage = imread(thisfile);
   [~, basename, ~] = fileparts(thisfile);   %just the name with no folder or extension

[Walter Roberson]

No, my code shows as far as extracting the basic filename (no folder, no extension) into the variable named basename but does not show creating a variable that is named the same as the content of the variable named basename . So if basename contained IMG001 then the user would want variable named IMG001 created, but doing that would involve either user some form of eval() or else generating some code into a file and executing the code. And of course if basename contained 001IMG because the file name was 001IMG.png then you simply cannot create a variable named 001IMG -- which should tend to suggest that this is not a wise thing to want to do.

[Image Analyst]

@hmhuang, no. He did not create the dynamically named variable. His variable names are hard coded. If the file was called foo.png, there is no variable created called foo. You could do that with eval() but as we've been discussing, that's not recommended so that's why he probably did not show it. However he's saying that some people think they want to do it because they don't realize the problems it can cause.

[hmhuang]

That's clear now, thanks!