Hi,

On 08/16/2014 08:10 AM, Clarkson, Matt wrote:
> Thanks Sascha.
>
> Quick feedback:
>    1. Our CLIs all run via a script. So I simply disabled auto-loading for 
> these modules, and removed some other debug from our code. This wasn’t a 
> problem before, in MITK-2013.09, as mitkCore didnt autoload lots of other 
> stuff. But I like the autoloading, it has enabled us to tidy up some other 
> code. i.e. unit tests don’t explicitly need to register factories, etc.etc. 
> So we have a workaround for now until 17581 is done.

Okay, we are working on it.

>    2. It turned out to not be an LD_LIBRARY_PATH problem. Well…actually, it 
> was 2 things. We were running external apps like NiftyReg, which I needed to 
> recompile, as I was picking up an incorrect library which then had a 
> dependency i was unaware of. However, I also found lots of other problems in 
> the XML. In the last round of CLI changes, we made it so that all errors are 
> caught and a message box generated. This was not the case before, so things 
> that failed XML validation simply silently disappeared and never showed up in 
> the CLI plugin. Now we get an error message. So there was actual XML i needed 
> to fix to get past the strict validation.

The error reporting is paying off now :-)

> So, nothing for you to do.

Good to know, thanks for the update.

- Sascha


> Thanks
>
> M
>
> On 7 Aug 2014, at 12:14, Sascha Zelzer <[email protected]> wrote:
>
>> Hi,
>>
>> On 08/07/2014 10:47 AM, Clarkson, Matt wrote:
>>> Hi there,
>>>
>>> I just noticed the following after our recent MITK upgrade to 2014-03.
>>>
>>> 1. When a command line module is run, there can be lots of output on the 
>>> console std::out output to do with auto-loading of modules. Is there a way 
>>> to turn the output off, as it won’t be valid XML.
>> In this moment, some of our guys are working on exactly that issue. This is 
>> bug http://bugs.mitk.org/show_bug.cgi?id=17581 . If everything goes well, 
>> this should be configurable soon.
>>
>>> 2. It appears (although I may have diagnosed it wrong), that command line 
>>> apps no longer have the library path set to be that of the calling 
>>> application. For example, when run on Linux, MITK provides bash scripts to 
>>> setup the LD_LIBRARY_PATH to be the bin folder plus bin/plugins. This now 
>>> appears to launch the main app ok, but command line modules are not run, as 
>>> they dont have the same path.
>>>
>>> Is that right?
>> This is new to me. When starting MitkWorkbench from for the build directory, 
>> I don't use shell scripts but the few cmd line apps I am using just worked 
>> out of the box. The cmd line app process should actually inherit the 
>> environment from the parent process.
>>
>> It could also have to do with missing rpath information for specific cmd 
>> line apps, so their dependencies can not be resolved by the dynamic linker 
>> without extending LD_LIBRARY_PATH.
>>
>> Best,
>> Sascha
>>
>>


------------------------------------------------------------------------------
Slashdot TV.  
Video for Nerds.  Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users

Reply via email to