Merge directives: current state

Aaron Bentley at
Thu Mar 1 15:03:22 GMT 2007

Hash: SHA1

John Arbash Meinel wrote:
>> The top comment area is patch/email-armoured RIO format.  It's 76
>> characters wide, and preserves \r and \n.  Next, I'll deal with
>> trailing-space munging, but that's the last thing I have in mind.
> Interesting. Just FYI the default in Mozilla clients is 72 chars. (Which
> is why you see everything wrapped below). I don't strictly care about
> what wrapping we use. But if we are going to wrap maybe something shorter?

Sure, it ought to be an easy change.

> Also, do you wrap all lines of content, or just the header?

The patch or bundle is treated as a blob-- it's up to the patch/bundle
format to handle its own wrapping issues.

> Oh, and what happens if you have a string that ends in '\'. I don't see
> how any of the strings right now would have it. But hypothetically do
> you handle:
> # entry: foo\
> Where "foo\" is the actual value?

If "foo\" is the actual value, it should be

# entry: foo\\

> v- In your comments you mentioned including the list of gpg signatures
> as part of the message header. So you need to somehow know what
> signatures will be used, right?

Yes, but I'm considering leaving this functionality out for now, since
we don't have a way of verifying that a directive was signed with a
particular key.

> For my personal aesthetics, it would be nicer if the wrapped lines were
> indented. So that rather than:

It may be possible, but it gets a bit wacky, because RIO itself does
indenting, but it does it for multi-line values.  I'll ponder.

> I would recommend switching to a millisecond timestamp, rather than the
> full 9 decimal (nanosecond) timestamp. I don't know what timestamp this
> is (I assume it is new).

It's from the bundle serializer.  Actually, it's hard to see anything
more precise than the date being useful.

Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Mozilla -


More information about the bazaar mailing list