Comment 91 for bug 61463

AIUI, the change was effected mainly for the substantial execution-time benefits it brings, which are especially considerable for boot-time. "Making a point" was absolutely, certainly not one of the reasons for the change, and suggesting so is silly.

OTOH, I think some more extensive verification, before the switch, of dash's claims to be as POSIX-compliant as possible would have been nice. Personally, I'm miffed by dash's very poor and currently non-conformant implementation of arithmetic expansion, which doesn't handle identifiers properly, among other things. It also lacks POSIX's interactive editing (vi-mode), but that's a far lesser concern.

Regardless: developers can not and should not commit to never breaking any thing in new releases. If you are concerned about production servers, you should of course not be rolling out new software on them before testing them to your satisfaction. Obviously, things should not be "broken" without some compensating benefit, and there is obviously disagreement as to whether the benefit was, indeed, sufficiently compensatory. However, it is arrogance to claim that there /was/ no appreciable benefit, or that it is clear that the benefit was insufficient.

In any case, the decision /has/ been made, along with accompanying quite-large investments of development effort, so I don't see what point is served by complaining about it now, especially on a bug-tracker that is not intended for griping that you don't like how a particular decision had gone.