Hi,
I am running the Windows version of Squeak 2.4b on my laptop pentium processor using Windows95. I have gotten the image running a few times. But it almost always comes up with the following "crash.dmp" stack dump.
It runs fine on my desktop system running Windows95.
Anyone got any ideas what the crash is about?
Peter William Lount peter@smalltalk.org http://www.smalltalk.org/versions/SqueakSmalltalk.html
=============== crash.dmp file ==============
Thu Apr 29 11:03:59 1999
Current byte code: 209 Primitive index: 264 Stack dump follows:
25955220 Dictionary>scanFor: 25955064 Set>findElementOrNil: 25954908 Dictionary>at:ifAbsent: 25955996 [] in SystemDictionary>send:toClassesNamedIn: 25954752 OrderedCollection>do: 25954596 SystemDictionary>send:toClassesNamedIn: 25954440 SystemDictionary>processStartUpList 25903356 SystemDictionary>snapshot:andQuit: 25903164 ScreenController>snapshotAndQuit 25902736 ScreenController>yellowButtonActivity 25902228 MouseMenuController>controlActivity 25902072 Controller>controlLoop 25901916 Controller>startUp 25902384 [] in ControlManager>activeController: 25902540 [] in BlockContext>newProcess
Peter -
I am running the Windows version of Squeak 2.4b on my laptop pentium processor using Windows95. I have gotten the image running a few times. But it almost always comes up with the following "crash.dmp" stack dump.
Your stack dump is identical to that of Pablo and Laurence, so clearly we have a systematic problem of some sort. All are Pentium machines running windows, and I believe that I recall your machine as being one of the ones with the "Float bug" that was noted last February. Both Andreas and I are at a loss as to what may be causing this problem, but we suggest that *all three of you* try the following:
If you can run Squeak 2.3 on the machine that crashes, see if you can print the result of
10000000000 asFloat
It should be 1.0e10, obviously, but it might crash on that machine.
If it crashes then, on *another* machine, start up 2.4b and put comments around <primitive: 54> in Float>>timesTwoPower:, and store the image with that change. Then test whether the altered image will run properly on the machine that was failing to run 2.4b before.
It is possible that the startup code in 2.4b touches this hardware bug, and triggers the later problem by corrupting some Squeak object. Disabling primitive 54 could be an effective workaround.
Please respond directly to me and not to the Squeak list (I have kept this public in case anyone else is having this problem).
Thanks - Dan
squeak-dev@lists.squeakfoundation.org