I was able to repeatably cause the behavior by starting a new method and not providing the closing brace. Removing the closing brace also triggered the behavior. It happened every few seconds when reparsing (semantic) or updating font lock (as seen when a string is open and not completed). I had to have enough code in the buffer to be bigger than the screen size (to trigger scrolling).
In the attached file if I remove the closing brace of testit1() I see the spontaneous C-L behavior. A while back (26 Mar) there was also a message about getting a stack overflow in the messages buffer from semantic parsing (RE: jde glitches). When I created a new file to test out the spontaneous C-L behavior I also encountered the stack overflow problem. Saving the file, closing, and re-opening the file cured the problem. GNU Emacs 21.1 on Win2k, jde 2.2.9beta9.1, elib-1.0, eieio-0.17beta4, speedbar-0.14beta4, and semantic-1.4beta14 with semantic-imenu.el v 1.39.2.5 from CVS (to include Eric Ludlams patch) -- Craig Kelley -----Original Message----- From: Rory Molinari [mailto:[EMAIL PROTECTED]] Sent: Friday, April 12, 2002 3:46 PM To: [EMAIL PROTECTED] Subject: Re: spontaneous C-L Mark Abrams writes: > "Berndl, Klaus" wrote: > >> The ECB versions >= 1.60 contain code for better handling eshell. >> My first question: Do you use the eshell package in combination with >> ECB? > > No I am not using eshell, unless it is part of a package that I am not > aware of. There is no eshell.el in my lisp directories. > > Mark I am not using it either, and sometimes see the refresh problem. jde-version: 2.2.9beta9.1 semantic-version: 1.4beta14 eieio-version: 0.17beta4 I found myself having this problem yesterday, while editing a java class too large to post to the group. In trying (and failing - see below) to boil my class down to a small example I found the following: 1) The refresh occurred (repeatedly) while I was editing an unterminated string, which was followed by another string, like this (cursor shown with >|<): String string1 = "A test" + "case >|< String string2 = "Another" + "string"; 2) The autorefresh reliably goes away when I terminate the first string with another double quote. 3) I successively removed chunks of the original file, each time verifying that the auto-refresh still happened. This worked until I made a deletion (somewhat before the first string above) which stopped the autorefresh. Confusingly, undoing the deletion, returning the buffer to the previous state (in which the autorefresh occurred), did not make the autorefresh start up again. Reloading the file, from the last "broken" state, did not make the autorefresh start up again. 4) When it happens, the autofresh takes place regularly, at a frequency I have come to associate with the auto-reparse (semantic, I think) that takes place after changes in the buffer. I don't have hard evidence for it being a semantic issue, though, as I didn't see any messages in the *Messages* buffer. I have not included the file which causes this problem, as the problem does not occur when the file is simply loaded up. I am sorry that this report is not more helpful. Rory
TestCtrlL.java
Description: Binary data
