[Modules] [Q] How to deal with stray-modules?

danielv at netvision.net.il danielv at netvision.net.il
Tue May 14 21:42:01 UTC 2002


On submitting Celeste in a form that is loadable, I have made some
conceptual progress :-) - I don't have to submit modules! I'll just
submit changesets for the classes that made the design good enough, then
some snippets for module creation and moving classes around (duh, I
know). 

Still haven't made it run, but that's just work.

BTW, it seems like I'm supposed to make a subclass of ModuleRefactorer
for the moving around snippets, but
a. From a quick skim, it's not clear where my code is supposed to hook
in.
b. It sure isn't clear why that'd be more useful than just changeset
postscripts.

Actual on topic responses below...

Henrik Gedenryd <h.gedenryd at open.ac.uk> wrote:
> danielv at netvision.net.il wrote:
> 
> > ***
> > How do I make my class, once again, part of a living module?
> > ***
> 
> The only thing you can't do with this module is anything to do with reaching
> it via its path.
That does tell me a little, but right now I don't know what requires a
path and what doesn't. Why moving a class into a module requires no
path, but moving it out of a module does require a path, specifically,
is beyond my reach to guess. Maybe in the weekend I'll read the code. 

> If you e.g. open an inspector on the instance then you can send any message
> to it via 'self'. 'self move...' etc
I ('m pretty sure I) tried to move the class out of a stray module and
into a normal one without using paths, I had variable pointing at both,
and that didn't work (which I was referring to above).

> > It's a bug that classes can be moved to stray module, in fact, stray
> > modules should not cooperate in any changes to the system at all. I've
> > just sent a fix to this particular bug.
> 
> This is unrealistic. It would require essentially every method to contain
> this check.
Correct me if I'm wrong, at the moment I take this statement as meaning
that changesets adding extra checks will be looked upon with bemused
acceptance... I'll submit them for any code that bites me.

> > PS - I petition to move "save and quit" apart from "quit". Who's with
> > me?
> 
> I agree--I got bitten by the same thing yesterday. Could be that the menu
> morphs are so dang slow that you release the button to try again before they
> appear.
> 
> Just moving save and quit to under save would suffice.
> 
> The problem with "save as new version" is that I have to go thru them and
> figure out what they are. So I now prefer save as with a meaningful name.
> 
> H



More information about the Squeak-dev mailing list