Peculiar behavior of lookfor

Given the following function definition

function is=isleapyr(yr)
%  returns T for input date being a leapyear

is=(datenum(yr+1,1,1)-datenum(yr,1,1))==366;

Why the following output from lookfor?

>> lookfor leap
isleapyr                       - returns T for input date being a leapyear
isleapyr                       - returns T for input date being a leapyear
>> help isleapyr
 returns T for input date being a leapyear

>>

That is, where does the doubled listing come from? I've ensured there's only one copy of isleapyr.m extant in the entire Matlab installation directory. It's annoying and apparently some sort of a bug???? OBTW, R2012b.

Well, the latter comment got me thinking -- so I copied the file to a working directory in my old R12 release and voila!!

>> lookfor leap
isleapyr.m: %  returns T for input date being a leapyear
>> isleapyr(2004)
ans =
   1
>> which isleapyr
C:\matlabR12\work\isleapyr.m
>>

So it appears to be some sort of introduced bug. Anybody else confirm still extant before I submit official report?

[Matt J]

What happens when you do

 >>which -all isleapyr

in your original setup?

[dpb]

>> which -all isleapyr
c:\ML_R2012b\work\isleapyr.m

>> ver
MATLAB Version: 8.0.0.783 (R2012b)
...
Operating System: Microsoft Windows XP Version 5.1 (Build 2600: Service Pack 3)
Java Version: Java 1.6.0_17-b04 with Sun Microsystems Inc. Java HotSpot(TM) Client VM mixed mode
...
MATLAB                                                Version 8.0        (R2012b)
Curve Fitting Toolbox                                 Version 3.3        (R2012b)
Optimization Toolbox                                  Version 6.2.1      (R2012b)
Signal Processing Toolbox                             Version 6.18       (R2012b)
Statistics Toolbox                                    Version 8.1        (R2012b)
>>

Just in case somebody compares versions/toolboxen/etc....

[David Young]

I get the double listing too, in 2013b:

>> lookfor harris-stephens
corners_hs                     - Simple implementation of the Harris-Stephens corner detector.
corners_hs                     - Simple implementation of the Harris-Stephens corner detector.
>> which -all corners_hs
C:\Users\davidy\Documents\MATLAB\corners_hs.m

but it doesn't seem to happen for every file, even if they're in the same directory as this one, and I haven't figured out what the difference is.

[dpb]

OK, that's enough for me to be convinced they've introduced a bug--I'll submit a bug report.

Guess if anybody else can figure out the problem or a workaround they can add it here.

[David Young]

Oh maybe it is the same for all files in the directory. It looks like what's happening is that it's looking at the current directory, and then also at the directories on the path, including possibly the current directory.

[dpb]

I think your last supposition is correct...I hadn't noticed it was everything in the current directory with the problem previously. I just submitted the bug report; I referenced/linked to the thread here so if TMW will read it they'll see the comment. Of course, it shouldn't take long for them to uncover the recursion problem somebody introduced.

ADDENDUM

Just confirmed your hypothesis--moving out of the current directory causes the symptom to disappear.

[A Jenkins]

I get the same behavior on 2010a and 2008a as well, so I'm not sure we can say it was recently introduced...

Answer 1

[per isakson]

If the current directory is in the path lookfor seems to scan it twice

A little experiment in R2014a

    >> version
    ans =
    8.3.0.532 (R2014a)

With the current directory being the folder with isleapyr

>> lookfor leap
isleapyr                       - returns T for input date being a leapyear
isleapyr                       - returns T for input date being a leapyear
leapyear                       - Determine leap year.
decaydimer                     - Comparing SSA and Explicit Tau-Leaping Stochastic Solvers

and the current being another folder

>> lookfor leap
isleapyr                       - returns T for input date being a leapyear
leapyear                       - Determine leap year.
decaydimer                     - Comparing SSA and Explicit Tau-Leaping Stochastic Solvers

[dpb]

As noted earlier, I confirmed that with R2012b here as well.

It's now narrowed down to after R12 (ca 1999/2000) which works as expected here and R2008a which is the earliest another poster has documented to have the issue.

[per isakson]

"earlier" &nbsp I guess it took me too long experimenting and that you posted the comment during that period of time.

[dpb]

Looks roughly same time as D Young's observation but I didn't see yours until later...I'd not noticed earlier that it was only in the working directory that it occurs so that was a good catch by both of you.