Re: Mac OS X sscanf bug/feature?

[Ben Abbott]

On Mar 26, 2016, at 8:48 PM, Doug Stewart <address@hidden> wrote:

On Sat, Mar 26, 2016 at 8:18 PM, Ben Abbott <address@hidden> wrote:

Running the default branch on Mac OS 10.10.5, I get …

[val, count, msg, pos] = sscanf ("3I2", "%f")
val = [](0x1)
count = 0
msg =
pos =  4

Matlab returns …

[val, count, msg, pos] = sscanf ('3I2', '%f')
val =     3
count =     1
msg =     'Matching failure in format.'
pos =     2

I assume that Mac OS X is causing the problem. Can someone confirm that Octave’s sscanf relies upon a system library? i.e is this due to Mac OS X’s clang compiler/library, or due to something I should be able to fix.

Ben

p.s. I have 22 “make check” failures and am hoping to browse through them as time permits. This is first failure I’m looking at.

In ubuntu I get

 [val, count, msg, pos] = sscanf ("3I2", "%f")

val =  3

count =  1

msg =

pos =  2

--

DAS

Ok. Thanks.

I expect this is a Apple clang thing. Can anyone confirm?

Ben