[MINC-users] strange mnc2nii behaviour

[Andrew Janke]

> > > I Had mnc2nii crash on a long filename - here's and example (with a
> > > filename to illustrate the point):
> > > mnc2nii -signed -float t.mnc 0123456789012345678901234567890123456789.nii
> > > ... The usual header output
> > > MINC type 5 signed 1
> > > Segmentation fault

>  Yes it seems to only be a problem for that particular length. I am
> running on Linux but tested it on Mac today where it works without
> problems. Any ideas as to where to locate that error are appreciated.\

Hi Soren,

I had a bit of a quick hunt for this bug and I can say that it does
happen in MINC 2.0.9 but not in 2.0.13. What did change in between
these versions is the niftilib version.  However I am not entirely
sure that this is the problem.

What I do know however is that there is a limit to the filename size
that can be put into the nifti header (although I think that field is
no longer used in deference to the filename itself).

I think the limit was 14 ascii characters in the db_name field?

In any case the bug (Whatever it was) is squashed!  I like problems
that are fixed already, perhaps the answer is to "Backport" the 2.X
changes to 1.5. Although in this case it means little more than just
relinking against the new nifti libs (and one small change to nii2mnc
and mnc2nii IIRC).


a