On Fri, Jun 21, 2019 at 03:43:57PM -0500, Derek Martin wrote:
But regardless, it does beg the question whether, after about 25 years of no one bringing it up, we really need to consider making a change here. I'm not necessarily opposed but as you know, I generally favor a policy of "make changes conservatively, measuring the utility of the change against the risk of chainging things."
I'm poking around the code right now, and I've already found an argument against replacing *all* "-" with "_". The same function is also used for the output of %{} (e.g. %{charset}), and %t (content type). Both of these can have "-" in them.
The issue, though, is that the filename isn't always under the user's control. It has been a very long time without issue, but is there a possibility of program argument abuse that could lead to a security issue here?
-- Kevin J. McCarthy GPG Fingerprint: 8975 A9B3 3AA3 7910 385C 5308 ADEF 7684 8031 6BDA
signature.asc
Description: PGP signature
