So 3.6M1 would be containing the fix, anyone tried it?

J-D

On Thu, Jul 30, 2009 at 2:17 PM, Lars George<[email protected]> wrote:
> Hmmm,
>
> Looking at the JDT core release note I found this:
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=267046
>
> Seems like that is it.
>
> Lars
>
>
> Lars George wrote:
>>
>> Hi,
>>
>> I spent some time last night checking into this. The Eclipse stack dump
>> revealed that it goes apparently into an endless loop trying to get the
>> parameter names for a method in the Scan class. Here the abbreviated stack
>> dump:
>>
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858] "main" prio=6
>> tid=0x11001590 nid=0xa0855720 runnable [0xbfffc000..0xbffff798]
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.core.SourceMapper.getUnqualifiedTypeSignature(SourceMapper.java:1146)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.core.SourceMapper.getUnqualifiedTypeSignature(SourceMapper.java:1131)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.core.SourceMapper.getUnqualifiedMethodHandle(SourceMapper.java:1059)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.core.SourceMapper.getMethodParameterNames(SourceMapper.java:974)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.core.BinaryMethod.getParameterNames(BinaryMethod.java:160)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.internal.codeassist.InternalCompletionProposal.findMethodParameterNames(InternalCompletionProposal.java:120)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.core.CompletionProposal.findParameterNames(CompletionProposal.java:1950)
>> 7/27/09 8:13:37 PM [0x0-0x342342].org.eclipse.eclipse[17858]  at
>> org.eclipse.jdt.ui.text.java.CompletionProposalLabelProvider.appendUnboundedParameterList(CompletionProposalLabelProvider.java:106)
>> ...
>>
>> So a method in Scan is causing this. While testing I notriced that I can
>> open the completion window and scroll down until the readFields() method,
>> going down one more line it hangs. The next method is getFamilyMap(...). So
>> I started testing the following things, always recompiling the HBase jar and
>> restarting Eclipse:
>>
>> - Remove all JavaDoc
>>
>> Does not help.
>>
>> - Remove Scan return type from getFamilyMap()
>>
>> Does not help.
>>
>> - Removed the whole getFamilyMap() method
>>
>> Works! No Eclipse lockup.
>>
>> - Added getFamilyMap() again and changed NavigableSet to TreeSet
>>
>> Does not help, Eclipse locks up again.
>>
>> - Removed the "familyMap" parameter from getFamilyMap()
>>
>> Works!
>>
>>
>> So the current status is that the type of the "familyMap" parameter is
>> causing Eclipse to lock up. I checked the source code for the SourceMapper
>> class that Eclipse uses and there is indeed a recursive call to
>> "getUnqualifiedTypeSignature" from itself. That seems to return (or else we
>> would see StackOverflowExceptions) but not end the surrounding while loop.
>> So that loop is running endlessly.
>>
>> Next step would be to debug the Eclipse class for example.
>>
>> Any thoughts so far?
>>
>> Lars
>>
>>
>> Andrew Purtell wrote:
>>>
>>> Interesting.
>>>
>>> This happens to me with HBase trunk and the client.Scan class, if auto-
>>> complete is triggered. I'll try the workaround suggested, because I've
>>> had to resort several times to plain text editor, which is doable, but
>>> not convenient.
>>>
>>>   - Andy
>>>
>>>
>>>
>>>
>>> ________________________________
>>> From: Dave Latham <[email protected]>
>>> To: [email protected]
>>> Sent: Wednesday, July 1, 2009 7:05:21 AM
>>> Subject: Re: Eclipse crashes when developing Row Filters
>>>
>>> Hi Ramesh,
>>>
>>> I've had the same problem with Eclipse 3.3 and 3.4.  I'm hoping that the
>>> just released 3.5 will have it fixed.  It always seems to happen when
>>> Eclipse tries to auto-complete something from the filter interface, or if
>>> you try to open up the interface.  I've tested it with the prebuilt jars
>>> (0.17, 0.18, and 0.19), and jars I've compiled myself, with or without a
>>> source attachment, and it happens reliably.  The only solution I've found
>>> is
>>> to set up an eclipse project with the HBase source and include that
>>> during
>>> development of code that uses the filters.  Once I'm done with the
>>> filters,
>>> I switch back to using the jars.
>>>
>>> I'd love hear if someone else has found a better solution.
>>>
>>> Dave
>>>
>>> On Wed, Jul 1, 2009 at 5:35 AM, Ramesh.Ramasamy
>>> <[email protected]>wrote:
>>>
>>>
>>>>
>>>> Hi,
>>>>
>>>> I am using Eclipse 3.3, JDK 1.6.0_12 and Hadoop/Hbase 0.19.1.
>>>>
>>>> On coding using some of the filter classes, eclipse hangs, and have no
>>>> other
>>>> option to continue it unless kill/restart the process. Does any body
>>>> figured
>>>> it out the problem and have a fix?
>>>>
>>>> TIA,
>>>> Ramesh
>>>>
>>>>
>>>> --
>>>> View this message in context:
>>>> http://www.nabble.com/help-with-map-reduce-tp22925481p24289040.html
>>>> Sent from the HBase User mailing list archive at Nabble.com.
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>

Reply via email to