|
|
Hi Folks -
I have now repeatedly noticed that Monticello can get into deep trouble
when it ever displays "Reloading..." when loading/merging a package.
Concretely, I have seen errors which are due to (as yet) undefined
superclasses when trying to create a subclass.
Digging into it, it seems that all definitions that raise errors in a
first loading pass, are then subjected to a second load attempt (the
dreaded "Reloading..."). This raises two questions:
1) Looking at the places generating the errors in the first pass, one
can see that this is often triggered by a subclass being defined before
its superclass. But shouldn't the load order prevent that? (right now it
doesn't seem to, or maybe it's getting confused but I have various
packages in which I get these errors repeatedly)
2) If a load fails in one pass, then what makes us assume that two
passes are enough to resolve the conflict? As a matter of fact, in the
cases in question, making the loader repeatedly try to load definitions
that had errors in a previous pass, makes my problems completely go away.
Mostly, I'm wondering if I'm missing something blatantly obvious (like
"oh, yeah, dependency handling was really broken in that version of MC")
or if other people have ever encountered problems like these.
Cheers,
- Andreas
|
Locked
