[Insight-users] Insight Journal submission: problems at every step of the way.

Zachary Pincus zpincus at stanford.edu
Thu Feb 2 02:14:37 EST 2006


Julien -

Thanks for taking a look at these issues. I just uploaded a new  
revision that seems to take care of everything.

> Problem 2: you can upload the files as one archive or multiple  
> files, it shouldn't matter. Regarding the .tgz files I will take a  
> look, but they are supported.

I think it's the MIDAS system that doesn't recognize tgz files --  
it's not on their list of "supported file types".

> Problem 3. You should specify what ITK version you want to use.  
> Like ITK2.4.1 in your CMakeLists.txt.

Gotcha -- thanks for the advice. Would it be reasonable to have  
either (a) it be an error not to specify what ITK revision, or (b)  
have the default be the "latest-stable" revision?

I have also noticed that having to specify what version of ITK to use  
means that I can't build the project as-is on my system, because my  
system isn't set up with multiple versions of ITK -- I just have a  
single ITK tree, labeled 'ITK'. Trying to build with 'ITK2.4.1' as a  
required package results in a "Package not found" error. Worse, CMake  
then must be pointed at a directory with the file  
"ITK2.4.1Config.cmake" -- which doesn't exist on my system. So I  
can't even manually configure a build to work right. I just have to  
remember to change the CMakeLists.txt each time I change/upload the  
file.

I think that in general a solution that allows the same CMakeLists to  
work on both a normal developer machine and the special build  
environment would be best. I'm not sure of the best way to do this,  
though.

As to the specific build problem, it turns out that I was depending  
on the CVS version of ITK... I've fixed that. However, this raises  
another issue: if some IJ submissions are theoretically targeted at  
inclusion in ITK, they should be able to work with a recent CVS  
checkout (and include features therefrom). For example, my code was  
trying to throw an exception according to what seems to be the "new"  
best practices in the CVS. Targeting it for ITK2.4.1 means that it  
will need some exception handling changes if it is ever integrated  
into ITK -- or the exception throwing will be sub-optimal. If the  
build environment had a nightly, or weekly, CVS version of ITK to  
use, that might be useful.


> Problem 4. Sounds like a bug. I just fixed it. You should be able  
> to upload a new file, otherwise let me know.

Yep, it works fine now. However, I can't remove old bitstreams, nor  
update any of the other metadata in the MIDAS view (clicking 'update'  
just returns me to the original screen). So, for example, I can't re- 
designate my new PDF as the "primary bitstream" and I can't add  
useful comments to the newly-uploaded bitstreams. Thus, now there are  
some 4 or 5 bitstreams labeled "Source.tgz" and I can't remove the  
old ones or add information about which one is the current, working  
version.

> Just to let you know, the Insight Journal was implemented in a  
> short amount of time and we know that it is not optimal. We are  
> currently in the process of improving it, especially the submission  
> of new revisions (via cvs/svn) and the documentation. It should be  
> better in the near future.

Thanks for your work -- both in fixing these issues, and in getting  
the IJ off the ground to begin with. I was feeling a little down  
after being unable to get anything to work right, but I am certainly  
a supporter of the concept, and I'll try to help with bug reports and  
journal submissions as best I can.

Zach




>
> Again, I'd like to thank you for improving the Insight Journal, let  
> me know if I can do anything to help.
>
> Julien
>
> Zachary Pincus wrote:
>> Hi folks,
>> I just submitted a brief Insight Journal submission describing a  
>> 2D image contour extraction filter that I wrote, which I hope will  
>> prove useful and merit inclusion in ITK. Unfortunately, I have had  
>> some problems with this entire process, culminating in my code not  
>> building because of some bizarre configuration issue on the build  
>> machine that I of course cannot diagnose, and then being unable to  
>> submit a revision to said code in the hopes of remotely diagnosing  
>> the problem.
>> Here is a list of the problems I have encountered today with the  
>> Insight Journal.
>> Problem 0 is the simple fact that documentation for this entire  
>> process is scattered across at least two wikis (the itk wiki, and  
>> the insight software colloquium wiki), with important information  
>> (like the latex and source code templates for submission) only  
>> available on the itk wiki. Figuring out what, precisely, I needed  
>> to do and how to do it took far too long.
>> Problem 1 was that despite the fact that at the end of the new  
>> submission process, I clicked "finish" and got an email saying  
>> that the submission had been accepted, the paper did not show up  
>> in the Insight Journal. Only after I clicked on "My Publications"  
>> and saw at the end of the line a "finish publication" did I find  
>> yet more steps to complete. I guess I clicked only on the "finish"  
>> button inside the MIDAS frame, but perhaps not on the finish  
>> button outside of the frame? Having two finish buttons to press is  
>> confusing and counterintuitive.
>> Problem 2 centers around including source files for the automated  
>> build system. Nowhere was it particularly well-explained that  
>> source code should be uploaded alongside the article file -- I  
>> spent some time searching for a specific entry point to upload  
>> source for the automated build program. Only on my third perusal  
>> of http://www.insightsoftwareconsortium.org/wiki/index.php/IJ- 
>> Testing-Environment did I notice that source should be uploaded as  
>> a secondary file and the build script will find it there.
>> So, I included a Source.tgz archive as a secondary file. However,  
>> tgz files are marked as unsupported and the upload process gave me  
>> trouble about that, despite the fact that the build system  
>> apparently had no problem opening it.
>> Problem 3 is that the code will not build on the test system. The  
>> problem is that the #define'd constant ITK_LOCATION (defined in  
>> itkMacro.h) is mysteriously unavailable to my code on the build  
>> machine, despite the fact that this constant is used throughout  
>> the ITK codebase (grep for it), and itkMacro.h should have been  
>> included properly. The same code compiles fine on my machine,  
>> where header files are properly included.
>> Problem 4 came when I tried to fix the build issue. On the "My  
>> Publications" link, there are two options: "modify revision" and  
>> "post a new revision." Surprisingly, "modify revision" in no way  
>> allows you to modify the revision. It just allows the exact same  
>> revision to be re-posted. How this is a modification is beyond me.  
>> The "post new revision" link then takes me to an extremely  
>> confusing frame with the directions to "Scroll down the frame and  
>> click on 'Add Bitstream'." When I did this and attempted to upload  
>> a new version of the PDF and source.tgz, I was provided with an  
>> AUTHORIZATION REQUIRED page because the operation was not  
>> permitted. So I can't add a new file, and so I can't even begin to  
>> try to diagnose why the build fails on the build machine.
>> All in all, this process has been extremely frustrating. Though I  
>> support the intention behind this all, the fact is that I've spent  
>> a day writing a "journal submission" describing an algorithm that  
>> is as old as the hills and whose implementation is straightforward  
>> (and exceptionally thoroughly documented in the code itself), and  
>> was then rewarded with getting to spend all evening fiddling  
>> around with the submit process. This is not a good use of my time.
>> Next time I write code for a basic, classic image processing  
>> method that ITK needs but does not have for some reason, I'll have  
>> to think twice about bothering to deal with all of this.
>> Zach
>> _______________________________________________
>> Insight-users mailing list
>> Insight-users at itk.org
>> http://www.itk.org/mailman/listinfo/insight-users
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users



More information about the Insight-users mailing list