[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Newsgroup Home]
|
[news.eclipse.modeling.mdt.uml2] Re: Strategy for loading commercial UML XMI 2.1 (MaicDraw)
|
Rafael,
Are you sure it wasn't a heap overflow error? That's what I'm getting.
The document actually gets successfully parsed, but gets hung up at the end
while trying to resolve forward references... I hope to have some time to
debug this further today.
Kenn
"Rafael Chaves" <rafael@xxxxxxxxxxxx> wrote in message
news:h07khg$632$1@xxxxxxxxxxxxxxxxxxxx
>I got a StackOverflowError before an OutOfMemoryError, so I am guessing the
>code is getting into an infinite recursion due to some oddness in that
>model.
>
> Kenn Hussey wrote:
>> Eike,
>>
>> So far my investigation suggests a bug in EMF, so I'm not sure the
>> problem is related to the size/scalability of the model itself...
>>
>> Kenn
>>
>> "Eike Stepper" <stepper@xxxxxxxxxx> wrote in message
>> news:h0632v$se8$2@xxxxxxxxxxxxxxxxxxxx
>>> Dawid, Kenn,
>>>
>>> And for what I know, NoMagic is currently working on a CDO-based new
>>> version of their team server. Since CDO allows for very scalable EMF
>>> models there will be potential for fewer OutOfMemoryExceptions (although
>>> I'm not sure if your particular OutOfMemoryException is related) ;-)
>>>
>>> Cheers
>>> /Eike
>>>
>>> ----
>>> http://thegordian.blogspot.com
>>> http://twitter.com/eikestepper
>>>
>>>
>>>
>>> Kenn Hussey schrieb:
>>>> Dawid,
>>>>
>>>> I have successfully loaded XMI files produced by MagicDraw on several
>>>> occasions in the past, so I suspect this is more an issue with the
>>>> particular file/model than MagicDraw's serialization in general. FWIW,
>>>> I've
>>>> heard they're working on a version of their tool that's based directly
>>>> on
>>>> Eclipse UML2, so this shouldn't be an issue on the long term (but of
>>>> course,
>>>> I realize that doesn't help you with your current problem).
>>>>
>>>> We'll take a look at your sample model and try to determine what's
>>>> going on.
>>>>
>>>> Kenn
>>>>
>>>> "Dawid Loubser" <dawidl@xxxxxxxxxxx> wrote in message
>>>> news:h03dv1$kj1$1@xxxxxxxxxxxxxxxxxxxx
>>>>
>>>>> Good day,
>>>>>
>>>>> We are busy building a toolset which requires the loading of
>>>>> XMI-serialised UML produced by commercial tools (for now, we are
>>>>> focusing on MagicDraw UML).
>>>>>
>>>>> We have tried (unsuccessfully) for weeks to accomplish this, and are
>>>>> at
>>>>> our wits' end - please help!
>>>>>
>>>>> We cannot rely on the "export to EMF" functionality which happens to
>>>>> be
>>>>> provided by MagicDraw, as we need to support other tools in future,
>>>>> and
>>>>> are ultimately not in control of the files - we just want to access
>>>>> and
>>>>> transform the UML model elements in them.
>>>>>
>>>>> Of course, we need to develop this in a stand-alone fashion, but even
>>>>> so,
>>>>> the XMI file does not load in Eclipse via the EMF GUI tools either. I
>>>>> get
>>>>> anything from a "missing packages" error, to an "OutOfMemoryError"
>>>>> which
>>>>> brings down eclipse, or my stand-alone test program.
>>>>>
>>>>> Here is a standard XMI file containing two or so classes, can any
>>>>> expert
>>>>> here point us in the right direction to get this loaded? We are
>>>>> absolutely
>>>>> desperate at this stage, nothing we have tried seems to work. If we
>>>>> can
>>>>> load this file, we will be very happy:
>>>>>
>>>>> http://projects.solms.co.za/temp/testProject.xmi
>>>>>
>>>>> P.S. I am aware that there are a couple of dubious elements in this
>>>>> XMI
>>>>> file placed there by MagicDraw, but I would still expect EMF/UML2 to
>>>>> ignore elements it does not understand, not to wait 2 minutes and then
>>>>> fail with OutOfMemoryError. This is a tiny model, and I am running a
>>>>> VM
>>>>> which has 2GB of allocated memory.
>>>>>
>>>>> If there is a problem with this XMI file (produced by one fo the
>>>>> leading
>>>>> modeling tools as-is) any pointers as to the specific pre-processing
>>>>> we
>>>>> need to perform would be greatly appreciated.
>>>>>
>>>>> kind regards,
>>>>> Dawid Loubser
>>>>>
>>>>
>>>>
>>