Ambiguous -from of git-send-email

Hi, all

A trivial problem here, but I think it can be handled more clearly.

Recently when I begin to use git-send-email with my private email account, I found it didn't work as I expected. The problem is "From" of the patch. I have to specify my company email in the "From" which meant the author, but when I go with default value of git-send-email, I found it omit the additional "From" of patch in the mailbody.

For example, I specified "my@company.com" in patch's "From: " when used git-format-patch to generate it. And I use "my@home.com" to send the mail. But if I go with default setting of git-send-email, it said the mail would be sent as "from my@company.com"(of course it can't). At last, the sent mail only got "From my@home.com" in mail header, and "From my@company.com" was omitted.

I've checked the git-send-email code, and found what's the real meaning of "-from" specify in git-send-email is the email account(sender) rather than the author. And if the sender is the same as the author, the first line of patch "From: xxx" was omitted. But at least on my machine, the from specified in the mail would be overwritten by MTA/ISP(well, I am not sure which of them did this), so what's the content of "-from" is not important, but if "-from" is the same as "From" in the patch, the "From" line in the patch is omitted, otherwise the line is kept.

Yeah, it's reasonable. But I think it's not that clear. "-from" of git-send-email indicated the sender and "From" indicated the author is very ambiguous. Maybe use "-sender" here in parameter is more reasonable? And do we need some notes in manual?

And can we add a something like "sendemail.sender" to specify the sender? I think it's not that unusual to make author's email address is different from sender, at least I think it's more common than "sendmail.bcc".

If you agree, I'd like to offer a patch for "sendemail.sender"(maybe also give some note in manual for "-from").

Thanks!

-- regards Yang, Sheng