[MINC-development] "Amusing" mincblur issue...

Andrew Janke a.janke at gmail.com
Mon Nov 24 01:05:31 EST 2008


Hi all,

Try this on for size:

1. First add this to mincblur.c

@@ -192,6 +192,8 @@
     exit(EXIT_FAILURE);
   }

+  print ("FRED VERBOSE:%d CLOB:%d
GRAD:%d\n",verbose,clobber_flag,do_gradient_flag);
+
   /******************************************************************************/
   /* find the size of the blurring kernel, from one of -std, -fwhm,
-fwhm3d     */

2. Do this on a 32 bit machine:

mavis:mincblur$ ./mincblur -verbose -clobber ~/data/me/a.mnc /tmp/out.mnc
FRED VERBOSE:1 CLOB:1 GRAD:0
...

mavis:mincblur$ ./mincblur -clobber -verbose ~/data/me/a.mnc /tmp/out.mnc
FRED VERBOSE:1 CLOB:1 GRAD:0

...

3. Do the same on a 64 bit machine.

ben:mincblur$ mincblur -clobber -verbose ~/data/me/a.mnc /tmp/out.mnc
FRED VERBOSE:1 CLOB:1 GRAD:0
...

ben:mincblur$ mincblur -verbose -clobber ~/data/me/a.mnc /tmp/out.mnc
FRED VERBOSE:0 CLOB:1 GRAD:0
...

4. Scratch your head.

I will bet this is (yet another) 64bit problem with ParseArgv.{c,h}
but am open to suggestions...  I have only spent a bit of time trying
to hunt this down so far but it is beginning to get on my goat as it
means that in my case bestlinreg fails with an error that is somewhat
misleading.  ie:

   $mincblur -clobber -no_apodize -fwhm 8 -gradient
volgenmodel-work-flip/00/fit-model-iso.mnc
/tmp/bestlinreg-wL7Mdx7w/Sfit-model-iso_8
   File /tmp/bestlinreg-wL7Mdx7w/Sfit-model-iso_8_blur.mnc exists.
   Use -clobber to overwrite.
   Died at /data/home/rotor/opt/script_bin/bestlinreg line 280.

Of course here -clobber is given but in this case the -gradient option
is "knocking" it out.  I was first hunting for something wrong with
the -gradient option and then realised that the same problem happens
with -verbose and -clobber...

I will bet this same issue exists in other MINC tools too but have not
tried to find any yet, I had first thought it was an #define quoted
string problem as per what surfaced with rawtominc a while back that
Claude pointed out.

Thoughts?


-- 
Andrew Janke - andrew.janke at anu.edu.au
Department of Geriatric Medicine, ANU
(a.janke at gmail.com || http://a.janke.googlepages.com/)
Canberra->Australia    +61 (402) 700 883


More information about the MINC-development mailing list