@Nike – this sounds on the face of it to be a bug, since the behavior is indeed documented. I suggest that you contact MathWorks (support@mathworks.com) to report this issue. Please repost a comment here if and when you learn anything new.

xlsread('data.xlsx',1) 

used to read sheet 1 of data.xlsx

Now it gives:

[]

Now I have to do:

xlsread('data.xlsx','nameOfSheet1') 

In order to make it work.

Also undocumented, as the documentation says SHEET can still be a number.
Am I the only one with this problem ?

Also, why does it say this comment was written on 29 October 2013, it is actually 26 July 2014 today.

@Michelle – thank you for taking the time to look into this, and for providing this update.

One has to admit that this sort of situation is indeed quite rare in Matlab. Apparently MathWorks’ standard workflow is able to prevent most similar occurrences.

We’ve had a chance to look into this issue some more to try to understand what happened. This behavior change was indeed fallout from the major work we did to greatly enhance the basic mode support in 12b, giving the ability to read xlsx files on Mac and Linux, and Windows machines without Excel (among other goodies). As he was digging through the code, the developer noticed a bug – that the trimming behavior (which wasn’t well specified in the help) was inconsistent. Leading column behavior was different from leading rows and trailing rows and columns.

In my view, the issue here is that the incompatible nature of the bug fix slipped through the cracks, so we didn’t follow our normal process of weighing the benefits and costs to users. More importantly, this meant that we did not document the change in a release note, leaving it to users to stumble across on their own. Thanks, as always, for making sure we were aware of the impact on Andreas and potentially on additional users.

@Aurélien – This is the first I’m noticing your blog post. It seems related, but perhaps a different code change. I haven’t looked into it, so I don’t know what assessment we made about the change.

As I’ve said before, it can be a difficult balancing act between bug fixing and maintaining compatibility, and particularly in figuring out the impact (bad or good) of making changes. We try our best to do the right thing, and always appreciate when our users let us know when our change has impacted them in a way we might not have anticipated.

On 16 Dec 2011 , when testing the BETA R2012a I had contacted tech-support since this new behavior was not mentioned anywhere in the doc. Now my simple example 2 lines of 6 numeric values is even used in the doc : http://www.mathworks.fr/help/releases/R2012a/techdoc/rn/bs7oakc-1.html#bteesx_ lol!

As you imagin, we have a lot of codes which use importdata and which now require additional test for backward compatibility.
I definitely prefer MATLAB R2011* versions!!

(CORRECTION: obviously my real file didn’t contain 7,8,9 in the latter columns, but instead contained 4,5,6. Just didn’t want to muddy the water with that difference. The problem is in the “b” output)

R2009b gives:
>> [a,b] = xlsread('badcol.xlsx','Sheet1','')
a =
1 NaN NaN NaN
2 4 5 6
3 4 5 6
4 4 5 6
b =
'' 'a' 'b' 'c'

R2012b gives:
>> [a,b] = xlsread('badcol.xlsx','Sheet1','','basic')
a =
1 NaN NaN NaN
2 4 5 6
3 4 5 6
4 4 5 6
b =
'a' 'b' 'c'