Niko, What is the version of OCompletion in the image? I thought the fix in Ocompletion-RomainRobbes.36.mcz solved that issue. If you are indeed using that version, I'll have a look at your fix. Otherwise, it is just a matter of time before the latest version of OCompletion is used instead of the older one. Cheers, Romain On Oct 27, 2009, at 3:07 PM, Niko Schwarz wrote:
Hi list,
in the most recent dev image (9.10.5), OCompletion is broken for me.
I lost my reproducible test case, but the gist was that sometimes when OCompletion had no idea which type a temporal variable would be, you were unable to insert code behind that variable, because if you tried, a debugger would show up.
The problem appears to be Debugger>guessTypeForName:. debuggerMap can be nil and if it is, the method fails. I changed the method to the following, and then the strange errors that occur on my image disappear:
guessTypeForName: aString | index object | index := debuggerMap ifNotNil: [(debuggerMap tempNamesForContext: self selectedContext) indexOf: aString ifAbsent: []]. object := index ifNil: [index := self receiver class allInstVarNames indexOf: aString ifAbsent: []. index ifNil: [^ nil]. self receiver instVarAt: index] ifNotNil: [self selectedContext tempAt: index]. ^ object class
Essentially, I leave index at nil, if debugerMap is nil.
I don't know enough about OCompletion to tell whether this is a fix, or makes things worse somehow.
Cheers,
Niko
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
-- Romain Robbes http://www.inf.unisi.ch/phd/robbes