On 2016-08-03 12:36:06 +0200, Michael Tatge wrote: > Hi, > > * On Wed, Jul 27, 2016 02:46AM +0200 Vincent Lefevre ([email protected]) > muttered: > > Mutt should store the message to the Sent folder *after* running > > sendmail and *only* if sendmail returned with a zero exit status. > > NO, this was what mutt was doing long ago. The current behaviour was > introduced exactly to *prevent* the situation where sending fails and > you don't have a local copy. > Mutt should try to store the message first and *then* try to send it. > Mutt should not store the same message over and over, this is where > action needs to be taken.
This is not the only problem. If the message cannot be sent (nonzero exit status from sendmail), then this would mean that the message needs to be removed from the Sent folder. Otherwise the user may think that he sent the message while he didn't. > The concept of store before send is the right way to do it(tm) given the > current code. > Imagine you have a typo in the recipient's email. You send it, which > succeeds and then storing a (local or remote) copy fails for whatever > reason. What needs to be done, as I've explained, is to allow the user to provide another location to store the message. > You'll receive the bounce later realizing your typo but don't have a > copy to resend the message from. -> BAD The copy is in the other location provided by the user. And if it is not possible to store the message anywhere in a safe location, then with the "store first" method, this would mean that it is not possible to send the message, which can be even worse. -- Vincent Lefèvre <[email protected]> - Web: <https://www.vinc17.net/> 100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/> Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)
