On Thu, 10 Jun 2010 08:31:05 -0700, Carl Worth <cworth@xxxxxxxxxx> wrote: > I suppose we could maintain compatibility with any scripts, etc. by > still emitting the initial "From " line, but declaring these files as > messages (not mbox) and avoiding doing any quoting for them. > > I think that gets us all the upsides with no downsides. I'll send one > last patch for that. Thinking about implementing and testing this, I realized that a file that looks like an mbox but isn't an mbox will confuse "git am" slightly. It will think that it should unquote any ">From " lines, but that would end up being the technically wrong thing to do since the lines aren't quoted. I'm not sure what to do here that would cause the least undesirable breakage. Ignore this problem? Emit a line that still contains anything that scripts might be looking for but that "git am" could key off of as "not actually an mbox"? I suppose we could put a magic timestamp there, but that feels pretty creepy and fragile. Another option would be to just emit RFC2822 messages unless the user passes an explicit option to format-patch (such as --mbox, which would be implied by --stdout). Then git would generate legitimate (unqoted) messages and legitimate (quoted) mbox files. I'd leave it to you to decide whether the --mbox option should be on by default or phased in with a warning or whatever. What do you think? -Carl -- carl.d.worth@xxxxxxxxx
Attachment:
pgpuDjHfdUkM9.pgp
Description: PGP signature