I get a "selector changed" error when tring to load update "1164projectFlapPref-sw " into 2.4b. Has anyone else encountered this? --dick peskin
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
At 5:34 PM -0700 5/1/99, Richard L. Peskin wrote:>
I get a "selector changed" error when tring to load update
"1164projectFlapPref-sw " into 2.4b. Has anyone else encountered this? --dick peskin
Dick,
I've tried to reproduce this error by updating while in an mvc project, updating while in a morphic project, updating while flaps are showing, updating while flaps are not showing, etc., but haven't encountered the error.
In trying to track down this problem, it would help to find out just which method was being recompiled (presumably some method in class Project) when the error happens, so, if you would, please try to load updates afresh and when you bump into the "selector changed" problem again, bring up the debugger, and take a look at the method which felt the error, which should be Behavior.recompile:from:, and see what the value of the "selector" and "oldClass" parameters are, and also evaluate "methodNode selector" in that method.
A stack dump (i.e. the .debug file) might also be useful, as well as any additional information you can provide that might help explain what's different between your environment and the environments of those who don't experience the problem.
With this additional info (and also reports from others who may run into the same problem) hopefully we'll get to the bottom of this.
-- Scott
Scott -- Here is some of the debug info you requested for update 1164. At 10:05 PM -0700 5/1/99, Scott.Wallace@disney.com wrote:
In trying to track down this problem, it would help to find out just which method was being recompiled (presumably some method in class Project) when the error happens, so, if you would, please try to load updates afresh and when you bump into the "selector changed" problem again, bring up the debugger, and take a look at the method which felt the error, which should be Behavior.recompile:from:, and see what the value of the "selector" and "oldClass" parameters are, and also evaluate "methodNode selector" in that method.
In Behavior.recompile:from: -- selector is resumeProcess oldClass is Project However, the method window for methodNode looks like some characters are missing. That is methodNode (on the first path where the selector changed error occurs) is: cess "comment stuff" <method details> Evaluating 'methodsNode selector' returns "cess" (which confirms the above).
If I manually try to fix this by adding in Pro (to get Process), I get a similar problem on the next pass through. The method window for methodNode starts with "erminate...." which I assume should be "terminate...."
These problems are consistently reproducible by update over the net. That is, the missing characters are being sent everytime. My environment is rather standard, a PowerMac 9500 with 604e processor running 8.1.
A stack dump (i.e. the .debug file) might also be useful, as well as any additional information you can provide that might help explain what's different between your environment and the environments of those who don't experience the problem.
I'm attaching this but it is not very useful.
Let me know what else I can do. --dick -------- Debug.log selector changed! 3 May 1999 12:46:53 pm Project class(Object)>>error: Project class(Behavior)>>recompile:from: [] in Project class(Behavior)>>compileAllFrom: MethodDictionary>>keysDo: Project class(Behavior)>>selectorsDo: Project class(Behavior)>>compileAllFrom: Project class(Class)>>compileAllFrom: Project class(ClassDescription)>>validateFrom:in:instanceVariableNames:methods:wasPresen t: Project class(Class)>>validateFrom:in:instanceVariableNames:methods:wasPresent: Metaclass>>name:inEnvironment:subclassOf:instanceVariableNames:variable:words:po inters:weak:classVariableNames:poolDictionaries:category:comment:changed: Model class(Class)>>subclass:instanceVariableNames:classVariableNames:poolDictionaries :category: UndefinedObject>>DoIt Compiler>>evaluate:in:to:notifying:ifFail: Compiler class>>evaluate:for:notifying:logged: Compiler class>>evaluate:for:logged: Compiler class>>evaluate:logged: [] in RWBinaryOrTextStream(ReadWriteStream)>>fileInAnnouncing: String>>displayProgressAt:from:to:during: RWBinaryOrTextStream(ReadWriteStream)>>fileInAnnouncing: ChangeSorter class>>newChangesFromStream:named: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: UndefinedObject>>ifNil: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: OrderedCollection>>do: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: CursorWithMask(Cursor)>>showWhile: Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: Utilities class>>readServerUpdatesSaveLocally:updateImage: Utilities class>>updateFromServer ScreenController>>absorbUpdatesFromServer ScreenController>>popUpMenuFor: ScreenController>>presentHelpMenu ScreenController>>yellowButtonActivity ScreenController>>redButtonActivity ScreenController(MouseMenuController)>>controlActivity ScreenController(Controller)>>controlLoop ScreenController(Controller)>>startUp [] in ControlManager>>activeController: [] in BlockContext>>newProcess
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
I get exactly the same result with a virgin 2.4b image on my Mac and on my NT workstation.
Personally I chose to "proceed" and had probably a couple of dozen errors before getting past 1164. For some funny reason, that image is extremely unstable... ;'P
William
"Richard L. Peskin" wrote:
Scott -- Here is some of the debug info you requested for update 1164. At 10:05 PM -0700 5/1/99, Scott.Wallace@disney.com wrote:
In trying to track down this problem, it would help to find out just which method was being recompiled (presumably some method in class Project) when the error happens, so, if you would, please try to load updates afresh and when you bump into the "selector changed" problem again, bring up the debugger, and take a look at the method which felt the error, which should be Behavior.recompile:from:, and see what the value of the "selector" and "oldClass" parameters are, and also evaluate "methodNode selector" in that method.
In Behavior.recompile:from: -- selector is resumeProcess oldClass is Project However, the method window for methodNode looks like some characters are missing. That is methodNode (on the first path where the selector changed error occurs) is: cess "comment stuff"
<method details> Evaluating 'methodsNode selector' returns "cess" (which confirms the above).
If I manually try to fix this by adding in Pro (to get Process), I get a similar problem on the next pass through. The method window for methodNode starts with "erminate...." which I assume should be "terminate...."
These problems are consistently reproducible by update over the net. That is, the missing characters are being sent everytime. My environment is rather standard, a PowerMac 9500 with 604e processor running 8.1.
A stack dump (i.e. the .debug file) might also be useful, as well as any additional information you can provide that might help explain what's different between your environment and the environments of those who don't experience the problem.
I'm attaching this but it is not very useful.
Let me know what else I can do.
--dick
Debug.log selector changed! 3 May 1999 12:46:53 pm Project class(Object)>>error: Project class(Behavior)>>recompile:from: [] in Project class(Behavior)>>compileAllFrom: MethodDictionary>>keysDo: Project class(Behavior)>>selectorsDo: Project class(Behavior)>>compileAllFrom: Project class(Class)>>compileAllFrom: Project class(ClassDescription)>>validateFrom:in:instanceVariableNames:methods:wasPresen t: Project class(Class)>>validateFrom:in:instanceVariableNames:methods:wasPresent: Metaclass>>name:inEnvironment:subclassOf:instanceVariableNames:variable:words:po inters:weak:classVariableNames:poolDictionaries:category:comment:changed: Model class(Class)>>subclass:instanceVariableNames:classVariableNames:poolDictionaries :category: UndefinedObject>>DoIt Compiler>>evaluate:in:to:notifying:ifFail: Compiler class>>evaluate:for:notifying:logged: Compiler class>>evaluate:for:logged: Compiler class>>evaluate:logged: [] in RWBinaryOrTextStream(ReadWriteStream)>>fileInAnnouncing: String>>displayProgressAt:from:to:during: RWBinaryOrTextStream(ReadWriteStream)>>fileInAnnouncing: ChangeSorter class>>newChangesFromStream:named: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: UndefinedObject>>ifNil: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: OrderedCollection>>do: [] in Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: CursorWithMask(Cursor)>>showWhile: Utilities class>>readServerUpdatesThrough:saveLocally:updateImage: Utilities class>>readServerUpdatesSaveLocally:updateImage: Utilities class>>updateFromServer ScreenController>>absorbUpdatesFromServer ScreenController>>popUpMenuFor: ScreenController>>presentHelpMenu ScreenController>>yellowButtonActivity ScreenController>>redButtonActivity ScreenController(MouseMenuController)>>controlActivity ScreenController(Controller)>>controlLoop ScreenController(Controller)>>startUp [] in ControlManager>>activeController: [] in BlockContext>>newProcess
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
Guys -
This is almost certainly a problem with your .changes file, most likely due to erroneous text conversion during download or unpacking. The selector-has-changed error is complaining during a recompilation, indicating that your prior source code is not in good shape. You can verify this by browsing to Process>>resumeProcess *before* attempting to update.
I have run the same update without error, and have verified that the update in question is identical on each of our redundant servers.
Hope this helps
- Dan
---------------------- The symp>Scott -- Here is some of the debug info you requested for update 1164.
At 10:05 PM -0700 5/1/99, Scott.Wallace@disney.com wrote:
In trying to track down this problem, it would help to find out just which method was being recompiled (presumably some method in class Project) when the error happens, so, if you would, please try to load updates afresh and when you bump into the "selector changed" problem again, bring up the debugger, and take a look at the method which felt the error, which should be Behavior.recompile:from:, and see what the value of the "selector" and "oldClass" parameters are, and also evaluate "methodNode selector" in that method.
In Behavior.recompile:from: -- selector is resumeProcess oldClass is Project However, the method window for methodNode looks like some characters are missing. That is methodNode (on the first path where the selector changed error occurs) is: cess "comment stuff"
<method details>
...
At 10:46 AM -0700 5/3/99, Dan Ingalls wrote:
Guys -
This is almost certainly a problem with your .changes file, most likely due to erroneous text conversion during download or unpacking. The selector-has-changed error is complaining during a recompilation, indicating that your prior source code is not in good shape. You can verify this by browsing to Process>>resumeProcess *before* attempting to update.
I have run the same update without error, and have verified that the update in question is identical on each of our redundant servers.
Just so we are all on the same wavelength, I downloaded 2.4b again. First I used the "packaged" MacBinary (Squeak2.4b.sea.bin). I got the same (update 1164) selector error again, even with the freshly downloaded .changes.
I then downloaded the uncompressed .changes file (Squeak2.4b.changes) directly from the archive. Using this copy of the .changes, I was able to update (including 1164) with no problem.
Don't have clue as to why. --dick
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
Since my Mac is at home, over lunch I fired up a FTP client on NT and made sure it was set to binary. That allowed me to get past the 1164 errors w/out even a burp.
Everything is running fine now - I'm curious, Richard, are you using Stuffit 5? And if so, what version? I've 5.02 at home and 5.1 on my Power Book which didn't have these problems.
William
"Richard L. Peskin" wrote:
At 10:46 AM -0700 5/3/99, Dan Ingalls wrote:
Guys -
This is almost certainly a problem with your .changes file, most likely due to erroneous text conversion during download or unpacking. The selector-has-changed error is complaining during a recompilation, indicating that your prior source code is not in good shape. You can verify this by browsing to Process>>resumeProcess *before* attempting to update.
I have run the same update without error, and have verified that the update in question is identical on each of our redundant servers.
Just so we are all on the same wavelength, I downloaded 2.4b again. First I used the "packaged" MacBinary (Squeak2.4b.sea.bin). I got the same (update 1164) selector error again, even with the freshly downloaded .changes.
I then downloaded the uncompressed .changes file (Squeak2.4b.changes) directly from the archive. Using this copy of the .changes, I was able to update (including 1164) with no problem.
Don't have clue as to why. --dick
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
If at any point the expansion of a compressed file invokes the use of Stuffit or Stuffit Expander, the preferences set in the application will determine how the EOL characters in the file are treated. This will sometimes happen even if the the file is ".sea" (self-extracting archive) if the browser settings are set up that way. I had similar "out-of-sync" errors on 2.4 until I beat Stuffit into submission.
Bottom line - check your Stuffit prefs.
At 10:46 AM -0700 5/3/99, Dan Ingalls wrote:
Guys -
This is almost certainly a problem with your .changes file, most likely due to erroneous text conversion during download or unpacking. The selector-has-changed error is complaining during a recompilation, indicating that your prior source code is not in good shape. You can verify this by browsing to Process>>resumeProcess *before* attempting to update.
I have run the same update without error, and have verified that the update in question is identical on each of our redundant servers.
Just so we are all on the same wavelength, I downloaded 2.4b again. First I used the "packaged" MacBinary (Squeak2.4b.sea.bin). I got the same (update 1164) selector error again, even with the freshly downloaded .changes.
I then downloaded the uncompressed .changes file (Squeak2.4b.changes) directly from the archive. Using this copy of the .changes, I was able to update (including 1164) with no problem.
Don't have clue as to why. --dick
================================= R. L. Peskin, Rutgers Univ. ;peskin@caip.rutgers.edu;http://www.caip.rutgers.edu/~peskin RLP Consulting rlpcon@sover.net; http://www.sover.net/~rlpcon
=================================================== Duane Maxwell dmaxwell (at) launchpados.com CTO http://www.launchpados.com Launchpad, Inc. (619) 578-8500 x226
Information contained herein is my personal opinion and not necessarily that of Launchpad, Inc. ===================================================
squeak-dev@lists.squeakfoundation.org
-
Dan Ingalls -
Duane Maxwell -
Richard L. Peskin -
Scott.Wallace＠disney.com -
William Barnett-Lewis