[Insight-users] App crashes when saving through itk::ImageSeriesWriter

alvaro numa barbeira abarbeira at fuesmen.edu.ar
Tue Aug 7 10:51:19 EDT 2007


Hello.

I´ve been experiencing problems with the saving process in the application
I´m developing, with ITK 3.0.1 and 3.2.

I use ITK in several registration modules (as well as doing IO), and FOX
for the GUI part. The main issue I´ve found is when calling
"ImageSeriesWriter::Update()", in Windows 2000 boxes, compiled with mingw
(see p.s.1).

Though I use it in a try/catch module, my application ocasionally crashes
without calling the code in the "catch" block. Depending on the service
pack, the OS presents a MessageBox accusing "The program has generated
errors and must be closed", or nothing at all. The "Update()" method
actually gets to generate some of the files before crashing my app, and
the last file generated has "0 bytes" size.

It has me at odds because there is no way to predict which sort of inputs
causes the conflict. Sometimes, the same imput gets saved with no problem
at all, and the next that same input causes a crash. Also if I try to
overwrite existing files, the program dumps.

In Linux, with gcc/g++, there are no such problems, my application behaves
exactly as I expect, so it is my guess the problem is not in my code.
¿Could it be the GDCM factory implementation in windows? ¿Or is it just
that the OS doesn´t like giving me space in the disk? I do nothing that is
not akin to the ITK Sotware Guide. I´ve found nothing suspicious to my
limited programming skills in the sources involved.

So, if anyone could please give me a clue...

Thanks in advance!


                    Alvaro


P.S.1: I know there are several issues compiling ITK with the latest
mingw. In the last ITK release, such as the definition of "boolean" in ITK
software conflicting with a header file ("rpcndr.h") in mingw distribution
(a quick bypass consists in altering the header´s "typedef" in a similar
way to ITK, and after building the library changing it back to its
original code, otherwise it will compile wrongly several mingw macros.
Tweaking the typedefs in ITK the other way around seems to be a never
ending task. Defining HAVE_BOOLEAN symbol just doesn´t get ITK to build).
ITK 3.0 compiles just fine.
       Migrating to cygwin is not an option for me (though I´ve managed to
build ITK and cmake after several hours of manually installin
necessary libraries), and VCPP is completely out of the questiom.
>:)

P.S.2: Though I have been working with ITK for several months (having read
the software guide, and sources), I´m still a newbie. However I´m
interested in taking part in some developing of the Library; but I´m quite
disoriented in how to get involved (working in Argentina and being very
far and unrelated to any organization part of the project). Anyone could
give me a hint or direction?

:D


More information about the Insight-users mailing list