Georg Bauhaus <[email protected]> writes: > I wonder if this case of adding another argument to a profile > is far less of a problem at Eurocontrol, that is, if indentation > is only momentarily not working?
That's a good point; how intelligent does the indentation need to be in the case of invalid code? If it gets control structures right, but everything else is either "indent at same level" or "indent at next level", would that be sufficient? Just for comparison, GPS handles the case of typing in an 'if' statement without using a skeleton first. I'll have to look in the code to see how it does that, but I'm guessing good error recovery in the parser. > If so, then there are cases when error recovery of completion via > (virtual or real) skeletons is useful and others, when it seems > overkill. Define the latter cases as those in which programmers > will see less of a problem if indentation does not work for > a moment (until the parser can catch up). In the long run, that distinction will disappear; people will always complain when something "ought to work" :). But it does provide a route for incremental improvement. -- -- Stephe _______________________________________________ Emacs-ada-mode mailing list [email protected] http://host114.hostmonster.com/mailman/listinfo/emacs-ada-mode_stephe-leake.org
