On Nov 7, 2012, at 6:31 PM, Michele Lanza wrote:
Well, today Roberto and I briefly looked into the filesystem stuff, which is a nice starting point, but I guess we'll have to implement now a full-blown fact extractor to fill in FAMIX
Yes why would this be magically here. And especially for which languages with Java and not scala and not VB6 and not C# and not C… You can take snippet of code and pass it to verveineJ We will evaluate that for our scenario (which works really well in Smalltalk but since this is in Smalltalk then you get bashed and not credible) we want to see if we can extract change objects from git (now we have them for Smalltalk from MC)
Pharo will probably move to git when it is ready so we will probably lose all our effort around MC but this is life.
and Hismo models. For now we're at a dead end because of some UFT-8 crap Roberto just sent an email about to the pharo-dev list. Can't believe in 2012 this thing still haunts people
Did you read the spec of UTF-8 for fun? I did it a bit just to understand and it is not simple. And this is not even complete :) Now roberto should also probably send a debugger stack so that people can see because may be this is zipping problem.
Typically when an MCZ file is corrupted the system tries to load it and gets confused. It should not be the case here it happens to us in the past.
Stef
Doing is always more difficult when you do it, but also rewarding.