[MERGE] left align log output if it only contains merge revisions

[Andrew Bennetts]

Aaron Bentley wrote:
[...]
> > I really don't see the point of the indent in this case.
> > That they are not mainline revisions is clear.
> > The patch is intended to tidy the output specifically in this case.
> > Just because that case may be a 1% corner case does not make the patch
> > any less valid.
> 
> Special casing is always a warning sign.  Sometimes it is necessary, but
> it often means that you haven't thought the general case through
> carefully enough.
> 
> But my point about 99% is that the 99% of cases build the expectation
> that the flush-aligned revisions are mainline revisions.

You could equally well argue that 99% of cases build the expectation that the
output of bzr log always starts unindented.

I think the concept that log of only non-mainline revisions is indented to
remind you that they aren't mainline is pretty subtle and is unlikely to be
understood by many, or even most, users.  If you feel strongly that the
indentation is still useful, what about displaying an unindented dummy log entry
first in this case:

"""
------------------------------------------------------------
(no mainline revision in range 2553.2.12..2553.2.13)
    ------------------------------------------------------------
    revno: 2553.2.12
    etc...
"""

But really, I think I'd rather not have screen real estate wasted on a useless
indent; as Kent says in the case of 2553.2.12..2553.2.13 it's clear they're not
mainline revisions.  I guess as a user I think about a revision range as
specifying "from here to here" on the revision graph, but actually with current
bzr log it's more like "from here to here via the mainline"?

If as a user I've asked to see the log of "revid:ABC..revid:XYZ", then I
probably don't care what the mainline of the current branch looks like.  I'm
trying to examine the history of XYZ, starting with ABC, i.e. treating XYZ as
the tip of a virtual mainline if you like.  (I know I've wanted to do this in
the past, and been frustrated that I get thousands of revisions rather than the
handful I was expecting, I hope Kent's fix for #4663 will help there.)

> Thanks.  I'm sorry if this has been frustrating for you.

Yeah.  Kent, thanks for tackling #4663, I'm looking forward to having that
fixed, regardless of the indenting :)

-Andrew.