On Fri, Apr 9, 2010 at 10:04 PM, Johan Corveleyn <jcorvel_at_gmail.com> wrote:
> On Fri, Apr 9, 2010 at 10:37 PM, Hyrum K. Wright
> <hyrum_wright_at_mail.utexas.edu> wrote:
> >
> >
> > On Fri, Apr 9, 2010 at 9:17 PM, Johan Corveleyn <jcorvel_at_gmail.com>
> wrote:
> >>
> >> Hi devs,
> >>
> >> Just a heads up, not really important, but I came across this in
> >> blame_cmd.c (inside print_line_info):
> >>
> >> [[[
> >> const char *rev_str = SVN_IS_VALID_REVNUM(revision)
> >> ? apr_psprintf(pool, "%6ld", revision)
> >> : " -";
> >>
> >> ]]]
> >>
> >> Which means that users of a repo with over 1 million revisions will
> >> see misaligned blame output ...
> >>
> >> So I guess this is coming up for you guys when s.a.o reaches the 1
> >> million mark :-).
> >
> > Heh. We also have a similar problem when committer names are longer than
> a
> > given threshold.
>
> Indeed, 10 characters:
> [[[
> return svn_stream_printf(out, pool, "%s %10s ", rev_str,
> author ? author : " -");
> ]]]
>
> > I've thought about hacking a solution into the client, but
> > have never really gotten around to it. Do you want to write a patch?
>
> Hm, yes. It might be a good opportunity for me to get my feet wet (I'm
> actually looking at making blame faster, cf. mail thread of a couple
> of weeks ago; but it might be a good idea to try something simpler
> first :-)).
>
> However, not having thought long about this problem: what would be the
> desired behavior?
>
> Should I first determine the maximum length of rev_str and author for
> the entire file and then use that as the column width (maybe with a
> minimum of 6 for rev_str and 10 for author, like it is now) ?
That would be my initial guess. Blame itself has enough overhead that
adding this calculation shouldn't be too difficult. It would also give you
a bit of exposure to the blame code itself. This may be overkill; I'll let
you do the exercise to find out. :)
-Hyrum
Received on 2010-04-09 23:15:12 CEST