On Wed, Jun 02, 2021 at 09:17:20AM +0200, Gerald Pfeifer wrote: > On Tue, 1 Jun 2021, Segher Boessenkool wrote: > > Brown paper bag time. The List-Id: should look like a hostname, not > > like an email address. Somehow I put in an at-sign when changing my > > gcc-patches example to the match-all example we have here. > > That's how things looked like before when using the Sender: header, > for example. So totally understandable.
And the existing example (like most procmail recipes) uses unescaped dot where a literal dot would perhaps be better. > > -* ^List-Id: .*<.*@gcc.gnu.org>$<br /> > > +* ^List-Id: .*<.*.gcc.gnu.org>$<br /> > > In my own filters I use > > ^List-Id: .*gcc.gnu.org > > to make it simpler and increase robustness around "<" and ">" (and "$"). And decrease robustness elsewhere (it now will match any list id that has the string gccXgnuYorg anywhere in it, where X and Y can be any character). > Or ^List-Id: .*gcc(-announce|-patches|-cvs-wwwdocs)?.gcc.gnu.org in > one case. > > What do you think? I only filter "gcc", "gcc-patches", "gcc-bugs" myself, all to separate folders. I use .*<...>$ on all lists I have everywhere, it works fine for me. It already isn't super strict, and more lax might work fine as well (there is absolutely nothing that prevents anyone from sending specially crafted emails to wreak havoc on everyone's filters anyway). If you change the example, maybe make it less peculiar at the same time? Filtering a high-traffic list like gcc-patches to the same folder as gcc-announce is a strange thing to do. (And please do make a change, so I don't have to brown-paper-bag a second time, this time for <, sigh). Segher
