J has a tendency to crash when I'm developing opengl code which uses locales.
These crashes have not been repeatable, but possible culprits include: (*) code I've written which has produced J errors, including errors in event handlers. (*) temporary locales (*) J's debugger (*) event handlers which display information using 1!:2&2 (*) my video drivers (*) windows xp (*) sms (which provides its own "mirror" video drivers on top of the drivers for the card) These crashes may not be specific to the beta, because I've seen similar crashes under j5.04 A possibly related issue is that sometimes the window close event handler will destroy the temporary locale without successfully closing the associated window (requiring wd'reset'). This has happened both with wd 'pclose' and wd 'psel parent; pclose' implementation of the close in the event handler (before codestroy). My most recent problems with wd 'psel parent; pclose' did not begin until I began using childHWNDC=:glsel 'child' in the constructor and glsel childHWNDC in the paint event handler (because sometimes the child control was not selected in the paint event handler). I've added this to my close event handler but won't know for some time whether this addresses the crashing problem -- unfortunately the crashes are not reliable. Another possibly related issue is that there are times when the window does not initialize immediately. I'll see huge cpu expenditures at window open time, and the entire display will blank and flash slightly at the rate of about one flash per second for something like six seconds before the operating system resumes its display. Process explorer (from sysinternals.com) indicates that most of the cpu is being expended by the system -- not by the application code, but it still attributes the expenditure to j. This kind of behavior (though not always the same cpu expenditure) happens when bringing an opengl window the the front and when bringing a non-opengl window to the front when an opengl window is being displayed. I've also seen this "flashing and slow" problem on another machine in contexts which did not involve J nor opengl, but did involve sms (there, the solution for this particular problem was to upgrade the video drivers). So this might be a red herring. I don't know how many J bugs are involved here, but it's at least zero. -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
