[BUG] Monticello problem while mucking with Parser

Ned Konz ned at bike-nomad.com
Wed Aug 6 02:54:28 UTC 2003


On Tuesday 05 August 2003 06:37 pm, Colin Putney wrote:
> As for Monticello, Avi and I have plans to make snapshot loading
> more atomic, in order to avoid just these sorts of issues. It looks
> like it will be fairly tricky, though, so it's hard to say when it
> will be done. If any one else gets bitten by this, do let me know.

I had a similar situation. Specifically, I defined a Morph class that 
had the following methods on the class side:

	#additionsToViewerCategories
	#additionsToViewerCategoriesGeometry
	#additionsToViewerCategoriesConnection

The first one called both of the others.

Unfortunately, the first one's name sorts before the others.

And when a method matching 'additionsToViewerCategories*' gets 
compiled, it gets run immediately...

which bombs out.

Another similar situation is where I modify a method in Morph to call 
another method that unfortunately gets loaded afterwards. Because of 
this, the progress notifications cause a walkback that can't be 
caught.

So I ended up having to package all of this in a SAR along with a 
carefully hand-edited .st file that defines stub methods that may get 
called during loading. I first load this .st file and then the .mcv 
file.

-- 
Ned Konz
http://bike-nomad.com
GPG key ID: BEEA7EFE



More information about the Squeak-dev mailing list