On Tue, Aug 23, 2016 at 12:18:42PM -0500, Derek Martin wrote: > Ultimately, all this is why I changed my mind and agreed with Kevin's > change.
Thanks Derek. It was pointed out to me off-list that I might have been a bit hasty to make a decision on re-creating a $locale alias, without giving a chance for the conversation to play out. Backward compatibility is tricky. On the one hand, I don't want to lightly cause massive breakage. On the other hand, because of the previous "C" default, there are a lot of examples and perhaps misinformation about the need to set $locale now. I think even keeping a $locale alias is still better with the changes made. $locale no longer affects the index or various other output, so even goofed-up, it causes much fewer problems. Perhaps we could create the alias but not document it. Or document it but mark it as deprecated, and delete it in e.g. 1.9. I'd appreciate more feedback/discussion and ideas from others about this change. One interesting thing Derek just mentioned is a conditional based on version. I know there is an ifdef patch, which sort of helps, but without an "else" doesn't completely solve the problem. -- Kevin J. McCarthy GPG Fingerprint: 8975 A9B3 3AA3 7910 385C 5308 ADEF 7684 8031 6BDA
signature.asc
Description: PGP signature
