Re: [Monotone-devel] speed of "mtn ls branches"

[Zack Weinberg]

On Jan 17, 2008 1:06 AM, William Uther
<address@hidden> wrote:
>
>
> On 17/01/2008, at 4:00 PM, Zack Weinberg wrote:
> > While working on something else today I noticed that the suspend-certs
> > implementation appears to be doing redundant work.  I'm not
> > remembering exact details, but it looked like: scan through all branch
> > heads and make a list of non-suspended ones, then check
> > --ignore-suspend-certs, then if it's not set, do that again.
...

I found the code I was referring to - it has nothing to do with "ls
branches", but I stand by my original statement.
update.cc, function calculate_update_set:

...
  bool have_non_suspended_rev = false;

  for (set<revision_id>::const_iterator it = candidates.begin(); it !=
candidates.end(); it++)
    {
      if (!app.get_project().revision_is_suspended_in_branch(*it, branch))
        {
          have_non_suspended_rev = true;
          break;
        }
    }
  if (!app.opts.ignore_suspend_certs && have_non_suspended_rev)
    {
      // remove all suspended revisions
      base64<cert_value> branch_encoded;
      encode_base64(cert_value(branch()), branch_encoded);
      suspended_in_branch s(app, branch_encoded);
      for(std::set<revision_id>::iterator it = candidates.begin(); it
!= candidates.end(); it++)
        if (s(*it))
          candidates.erase(*it);
    }
...

Seriously, tell me that's not doing the same thing twice.

zw