On 01Feb2013 21:20, Derek Martin <inva...@pizzashack.org> wrote: | On Fri, Feb 01, 2013 at 09:10:46PM -0600, Derek Martin wrote: | > On Sat, Feb 02, 2013 at 11:20:37AM +1100, Cameron Simpson wrote: | > > Given that they're both Maildirs on the same filesystem, why doesn't mutt | > > take this more efficient (both in time and disc space) approach? Just naive | > > use of an existing "save-message" library routine. | > | > I find your surprise to be very surprising. :) | > | > - Copy = actual copy... implies multiple physical copies, which you | > don't get with hard links | | Or, to say it another way (from the manual): | | <save-message> s save message/attachment to a mailbox/file | | ln != save-message. Even if all your mailboxes are maildir. But FWIW | one of the things I've been preaching for years is that Mutt's UI | should not behave differently based on the message store. If your | mail folders are mbox, making a hard link is completely nonsensical, | and mutt's behavior ideally should be consistent regardless of the | back end storage format.
ln doesn't change the UI behaviour. Just because mbox _forces_ a wasteful copy doesn't imply maildir should behave the same way. Speaking for myself, if messages have the same message-id I expect them to be essentially equivalent. I don't require them to be distinct files/copies. As remarked elsewhere, mutt's capable of a server-side IMAP move message. Nothing says whether that's a copy or a hard link behind the scenes. I agree there should be some user choice here; there may be people who rely on distinct copies. But they're not me, and it is an unnecessarily slow way to do things with maildirs. Cheers, -- Cameron Simpson <c...@zip.com.au> As you can see, unraveling even a small part of 'sendmail' can introduce more complexity than answers. - Brian Costales, _sendmail_