Hello, On Tue, Jul 26, 2016 at 09:01:38PM +0200, Danny Milosavljevic wrote: > Yeah, I thought about it but decided against it - there are very few options > in that file, substitute* can't substitute entire lines (or only at the > beginning of the line) or entire words (so it's not safe), the user is > supposed to set PREFIX and HASH (it's just a coincidence we didn't have to > change them) and if we did that then new versions of the package could sneak > in new options we wouldn't notice but we should have changed. Better for it > to fail instead of silently doing something strange.
well, substitute* is closer to a diff - it makes it easy to see what actually changes. And I would make the converse argument of you - for the next version, if more things change (an option is added to the file, for instance), then the substitute* still has a chance to get a working package. Imagine that someone without knowledge of the package might be the one to update it, for instance. I pushed with tiny changes to the description. Thanks! Andreas
