Class definition @ directory error

Hello,

I am writing a script wherein a new class 'polynomial' is defined. When I run the script, I get the following error:

 Error: File: polynomial.m Line: 49 Column: 10
A class definition must be in an "@" directory.

Upon typing which('polynomial'), I get the following output:

 /home/..../multipoly/@polynomial/polynomial.m

Clearly, polynomial.m is in a directory called @polynomial, but I'm still getting the error. I'm quite confused by this. I am running 64bit R2012a on Ubuntu 12.04.

Any help would be appreciated. Thanks.

Adi

[Walter Roberson]

Is multipoly your routine? Or is it (for example) inside Simulink ?

[Adi gahlawat]

It is my routine, I found the problem, kindly see me reply to Per Isakson below. Thanks for replying.

Answer 1

[per isakson]

I assume that

• you created the folder "/home/..../multipoly/@polynomial"
• you created the file "polynomial.m"
• the keyword, "class" appears in the file "polynomial.m" (line 49)

Do you try to use old or the new Class-Definition Syntax; do you use the new keyword classdef?

Did you try

    which('polynomial','-all')

[Adi gahlawat]

Hi Per,

the class definition and the folder name are all good. The problem was that on the path of the folder, there was another folder with '@' in its name that was throwing MATLAB off. I renamed the folder and now everything is fine. Thanks for replying.

Adi

[Jeremy Rutman]

I had a similar problem . It turns out ANY folder with ANY ampersand in it ANYWHERE along the path will munge things up royally. I officially declare this a bug

>> sm = segmodel( 'PROFILE', '0.16', 'use_real_pose', false );
Error using segmodel
Error: File: segmodel.m Line: 1 Column: 10
A class definition must be in an "@" directory.

>> which('segmodel','-all')
/home/jeremy/[e-mail]/dev/clothes_parsing/@segmodel/segmodel.m  
% segmodel constructor
>>

[Guillaume]

Well, giving syntactical meaning to filesystem paths was never a good idea in the first place, but I'm afraid Mathworks does not see it this way.

You can declare it a bug, but unless Mathworks completely overhaul their approach to paths, it's not going to change.