Home » Language IDEs » Java Development Tools (JDT) » 3.4 Hot Code Replace Failing
| 3.4 Hot Code Replace Failing [message #255395] |
Fri, 01 August 2008 16:47  |
Eclipse User |
|
|
|
Originally posted by: bcort.oculusinfo.com
I am finding that Eclipse 3.4 is much less effective at hot code
replacements while debugging than in version 3.3. For example, adding a
comment will cause it to prompt me to terminate/continue.
Are others also experiencing this?
Thanks
|
|
| | | | | |
| Re: 3.4 Hot Code Replace Failing [message #255588 is a reply to message #255584] |
Wed, 13 August 2008 11:48  |
| Eclipse User |
|
|
|
Originally posted by: bcort.oculusinfo.com
Sorry, that was unclear.
I tried adding a comment to running code and the reason given in the 'Hot
Code Replace Failed' dialog is "Scheme change not implemented".
Thanks
"B C" <bcort@oculusinfo.com> wrote in message
news:g7uv5i$k43$1@build.eclipse.org...
>I just tried adding a comment to running code and the reason is "Scheme
>change not implemented".
>
> Does this help diagnose?
>
> Thanks
>
> "Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
> news:g7o71e$49i$1@build.eclipse.org...
>> When the HCR fails, what exact error message are you getting, i.e. what
>> is the reason?
>>
>> Darins
>> http://www.runnerwhocodes.blogspot.com/
>>
>> "B C" <bcort@oculusinfo.com> wrote in message
>> news:g7fe6u$v2f$1@build.eclipse.org...
>>>I tried the simplest test case I could and hot code replace works fine.
>>>
>>> Is there any way to discover why hot code is failing in my production
>>> projects?
>>>
>>> The error log doesn't seem to show anything...
>>>
>>> "Walter Harley" <eclipse@cafewalter.com> wrote in message
>>> news:g70p5s$1q0$1@build.eclipse.org...
>>>>B C wrote:
>>>>> I am finding that Eclipse 3.4 is much less effective at hot code
>>>>> replacements while debugging than in version 3.3. For example, adding
>>>>> a comment will cause it to prompt me to terminate/continue.
>>>>>
>>>>> Are others also experiencing this?
>>>>>
>>>>> Thanks
>>>>
>>>> I've not had any problem with that, personally. If you can come up
>>>> with a reliable repro, with details of what JVM etc. you're using, I
>>>> think the JDT team would be very interested in a Bugzilla entry. Are
>>>> you sure that 3.3 vs 3.4 is the only thing that changed? Is this
>>>> happening in plain Java projects, or J2EE, or... ?
>>>
>>
>>
>>
>
|
|
|
| Re: 3.4 Hot Code Replace Failing [message #255605 is a reply to message #255588] |
Wed, 13 August 2008 12:49 |
| Eclipse User |
|
|
|
When HCR occurs, the Eclipse debug framework essentially fires off the new
bytes of the changed classfile(s) to the VM.
The VM may or may not accept them. In your case it is rejecting the
change(s).
Scheme change usual means you are adding methods or fields (or making
changes within an innerclass that results in synthetic method changes?).
What VM are you using? Possibly a different VM would
helpDarinshttp://runnerwhocodes.blogspot.com"B C" <bcort@oculusinfo.com>
wrote in message news:g7uvo4$rph$1@build.eclipse.org...
> Sorry, that was unclear.
>
> I tried adding a comment to running code and the reason given in the 'Hot
> Code Replace Failed' dialog is "Scheme change not implemented".
>
> Thanks
>
>
> "B C" <bcort@oculusinfo.com> wrote in message
> news:g7uv5i$k43$1@build.eclipse.org...
>>I just tried adding a comment to running code and the reason is "Scheme
>>change not implemented".
>>
>> Does this help diagnose?
>>
>> Thanks
>>
>> "Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
>> news:g7o71e$49i$1@build.eclipse.org...
>>> When the HCR fails, what exact error message are you getting, i.e. what
>>> is the reason?
>>>
>>> Darins
>>> http://www.runnerwhocodes.blogspot.com/
>>>
>>> "B C" <bcort@oculusinfo.com> wrote in message
>>> news:g7fe6u$v2f$1@build.eclipse.org...
>>>>I tried the simplest test case I could and hot code replace works fine.
>>>>
>>>> Is there any way to discover why hot code is failing in my production
>>>> projects?
>>>>
>>>> The error log doesn't seem to show anything...
>>>>
>>>> "Walter Harley" <eclipse@cafewalter.com> wrote in message
>>>> news:g70p5s$1q0$1@build.eclipse.org...
>>>>>B C wrote:
>>>>>> I am finding that Eclipse 3.4 is much less effective at hot code
>>>>>> replacements while debugging than in version 3.3. For example, adding
>>>>>> a comment will cause it to prompt me to terminate/continue.
>>>>>>
>>>>>> Are others also experiencing this?
>>>>>>
>>>>>> Thanks
>>>>>
>>>>> I've not had any problem with that, personally. If you can come up
>>>>> with a reliable repro, with details of what JVM etc. you're using, I
>>>>> think the JDT team would be very interested in a Bugzilla entry. Are
>>>>> you sure that 3.3 vs 3.4 is the only thing that changed? Is this
>>>>> happening in plain Java projects, or J2EE, or... ?
>>>>
>>>
>>>
>>>
>>
>
|
|
|
| Re: 3.4 Hot Code Replace Failing [message #255608 is a reply to message #255605] |
Wed, 13 August 2008 13:11 |
| Eclipse User |
|
|
|
Originally posted by: bcort.oculusinfo.com
I'm using Sun's JDK 1.6.0_06.
As I mentioned, this happens for inconsequential changes like adding
comments.
This problem does not happen with Eclipse 3.3 using the same VM.
"Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
news:g7v3l1$ejr$1@build.eclipse.org...
> When HCR occurs, the Eclipse debug framework essentially fires off the new
> bytes of the changed classfile(s) to the VM.
> The VM may or may not accept them. In your case it is rejecting the
> change(s).
>
> Scheme change usual means you are adding methods or fields (or making
> changes within an innerclass that results in synthetic method changes?).
>
> What VM are you using? Possibly a different VM would
> helpDarinshttp://runnerwhocodes.blogspot.com"B C" <bcort@oculusinfo.com>
> wrote in message news:g7uvo4$rph$1@build.eclipse.org...
>> Sorry, that was unclear.
>>
>> I tried adding a comment to running code and the reason given in the 'Hot
>> Code Replace Failed' dialog is "Scheme change not implemented".
>>
>> Thanks
>>
>>
>> "B C" <bcort@oculusinfo.com> wrote in message
>> news:g7uv5i$k43$1@build.eclipse.org...
>>>I just tried adding a comment to running code and the reason is "Scheme
>>>change not implemented".
>>>
>>> Does this help diagnose?
>>>
>>> Thanks
>>>
>>> "Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
>>> news:g7o71e$49i$1@build.eclipse.org...
>>>> When the HCR fails, what exact error message are you getting, i.e. what
>>>> is the reason?
>>>>
>>>> Darins
>>>> http://www.runnerwhocodes.blogspot.com/
>>>>
>>>> "B C" <bcort@oculusinfo.com> wrote in message
>>>> news:g7fe6u$v2f$1@build.eclipse.org...
>>>>>I tried the simplest test case I could and hot code replace works fine.
>>>>>
>>>>> Is there any way to discover why hot code is failing in my production
>>>>> projects?
>>>>>
>>>>> The error log doesn't seem to show anything...
>>>>>
>>>>> "Walter Harley" <eclipse@cafewalter.com> wrote in message
>>>>> news:g70p5s$1q0$1@build.eclipse.org...
>>>>>>B C wrote:
>>>>>>> I am finding that Eclipse 3.4 is much less effective at hot code
>>>>>>> replacements while debugging than in version 3.3. For example,
>>>>>>> adding a comment will cause it to prompt me to terminate/continue.
>>>>>>>
>>>>>>> Are others also experiencing this?
>>>>>>>
>>>>>>> Thanks
>>>>>>
>>>>>> I've not had any problem with that, personally. If you can come up
>>>>>> with a reliable repro, with details of what JVM etc. you're using, I
>>>>>> think the JDT team would be very interested in a Bugzilla entry. Are
>>>>>> you sure that 3.3 vs 3.4 is the only thing that changed? Is this
>>>>>> happening in plain Java projects, or J2EE, or... ?
>>>>>
>>>>
>>>>
>>>>
>>>
>>
>
>
|
|
|
| Re: 3.4 Hot Code Replace Failing [message #255612 is a reply to message #255608] |
Wed, 13 August 2008 13:07 |
| Eclipse User |
|
|
|
Time for a bug report then with a reproduceable test case.
Sorry I have not helped out
Darins
http://runnerwhocodes.blogspot.com
"B C" <bcort@oculusinfo.com> wrote in message
news:g7v4jg$go7$1@build.eclipse.org...
> I'm using Sun's JDK 1.6.0_06.
>
> As I mentioned, this happens for inconsequential changes like adding
> comments.
>
> This problem does not happen with Eclipse 3.3 using the same VM.
>
>
> "Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
> news:g7v3l1$ejr$1@build.eclipse.org...
>> When HCR occurs, the Eclipse debug framework essentially fires off the
>> new
>> bytes of the changed classfile(s) to the VM.
>> The VM may or may not accept them. In your case it is rejecting the
>> change(s).
>>
>> Scheme change usual means you are adding methods or fields (or making
>> changes within an innerclass that results in synthetic method changes?).
>>
>> What VM are you using? Possibly a different VM would
>> helpDarinshttp://runnerwhocodes.blogspot.com"B C" <bcort@oculusinfo.com>
>> wrote in message news:g7uvo4$rph$1@build.eclipse.org...
>>> Sorry, that was unclear.
>>>
>>> I tried adding a comment to running code and the reason given in the
>>> 'Hot Code Replace Failed' dialog is "Scheme change not implemented".
>>>
>>> Thanks
>>>
>>>
>>> "B C" <bcort@oculusinfo.com> wrote in message
>>> news:g7uv5i$k43$1@build.eclipse.org...
>>>>I just tried adding a comment to running code and the reason is "Scheme
>>>>change not implemented".
>>>>
>>>> Does this help diagnose?
>>>>
>>>> Thanks
>>>>
>>>> "Darin Swanson" <Darin_Swanson@us.ibm.com> wrote in message
>>>> news:g7o71e$49i$1@build.eclipse.org...
>>>>> When the HCR fails, what exact error message are you getting, i.e.
>>>>> what is the reason?
>>>>>
>>>>> Darins
>>>>> http://www.runnerwhocodes.blogspot.com/
>>>>>
>>>>> "B C" <bcort@oculusinfo.com> wrote in message
>>>>> news:g7fe6u$v2f$1@build.eclipse.org...
>>>>>>I tried the simplest test case I could and hot code replace works
>>>>>>fine.
>>>>>>
>>>>>> Is there any way to discover why hot code is failing in my production
>>>>>> projects?
>>>>>>
>>>>>> The error log doesn't seem to show anything...
>>>>>>
>>>>>> "Walter Harley" <eclipse@cafewalter.com> wrote in message
>>>>>> news:g70p5s$1q0$1@build.eclipse.org...
>>>>>>>B C wrote:
>>>>>>>> I am finding that Eclipse 3.4 is much less effective at hot code
>>>>>>>> replacements while debugging than in version 3.3. For example,
>>>>>>>> adding a comment will cause it to prompt me to terminate/continue.
>>>>>>>>
>>>>>>>> Are others also experiencing this?
>>>>>>>>
>>>>>>>> Thanks
>>>>>>>
>>>>>>> I've not had any problem with that, personally. If you can come up
>>>>>>> with a reliable repro, with details of what JVM etc. you're using, I
>>>>>>> think the JDT team would be very interested in a Bugzilla entry.
>>>>>>> Are you sure that 3.3 vs 3.4 is the only thing that changed? Is
>>>>>>> this happening in plain Java projects, or J2EE, or... ?
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>>
>
|
|
|
| Re: 3.4 Hot Code Replace Failing [message #256188 is a reply to message #255608] |
Mon, 08 September 2008 12:08 |
| Eclipse User |
|
|
|
I'd just wanted to chip in with a comment here. I stumbled upon this
thread while Googling for any help on a similar issue: I have also noticed
that the hot code replacement have become EXTREMELY picky about what
changes it accepts. Since this isn't based upon hard facts, the bottom
line is that I believe I had much more benefit of the HCR feature before
than after the 3.4 upgrade. (I've run -clean, but not made a new
workspace, to pre-empt comments from a later thread about the same
problem).
What I think often is happening, at least in my case, is that for
non-static inner classes, the hot code replacement fails upon
(user-perceived) minor changes such as accessing or not accessing anymore
a private field/method in the outer class. What's happening here, is that
the compiler has to insert an outer-class accessor method to access it, or
delete these accessor methods if they aren't needed anymore. I actually
believe both insertion and deletion worked earlier, but I am not certain
of the "insert" part. However, that it fails on the deletion is rather
annoying - couldn't it just have let that method be in place until next
restart? I am quite certain that it didn't crash on this specific thing
before.
These issues leads to the annoying fact that if you comment out some code
line in the inner class that includes an access to any private field or
method in the outer class, the HCR stuff fails. And this seems rather
strange, as one seemingly haven't done any "major changes" like overtly
adding fields or methods.
|
|
|
Goto Forum:
Current Time: Wed Jun 04 10:21:25 EDT 2025
Powered by FUDForum. Page generated in 0.32907 seconds |
Home