Re: [PATCH] git-send-email: die if sendmail.* config is set
- From
Drew DeVault <sir@cmpwn.com>
- Date
- Jul 18, 2020, 18:25 UTC
- Message-ID
- <C49YLQZUZ4RG.129ZY8FQEYDYU@homura>
- In-Reply-To
- <xmqqpn8sbt0w.fsf@gitster.c.googlers.com>
On Sat Jul 18, 2020 at 7:02 AM EDT, Junio C Hamano wrote:
Show 11 quoted lines
> It is not very friendly thing to do, though. > > It just closes the door for anybody to add something that works > independent from "git send-email", to which "sendmail.*" variables > may be appropriate knob to use. > > Demoting the "die" to "warn" is OK, and limiting the check to > variables that actually has corresponding and likely-misspelt > "sendemail.*" counterparts would be even better, but "you are not > allowed to have any 'sendmail.*' variables, ever" is way too much, > I am afraid.
I originally had this just a warning, but it can be difficult to see. There's a lot of text printed from git send-email, a lot of it looks like diagnostics, and it can be easy to lose a message in there. And if you pass --annotate, which I encourage people to do, your editor immediately covers up the warning.
I wonder if a happy medium would be adding a config option which squelches the message? If sendemail.squelchConfigError is set, for example. If not, I'll experiment with a noisy warning.