Fix confidence check and problems with successive builds due to symlinks.

Fix confidence check and problems with successive builds due to symlinks.

Set a parameter when checking "found" confidence so that it won't
abort with only two attempts yet made.  Also, switching versions can
leave behind symlinks that will break the next build so remove all
symlinks when switching.

BUG=

Committed: https://crrev.com/500be8ebfa734283910df8c30de133a219296601
Cr-Commit-Position: refs/heads/master@{#332434}

[bcwhite, 2015-06-01 18:22:28]

bcwhite@chromium.org changed reviewers:
+ simonhatch@chromium.org

[shatch, 2015-06-01 18:29:29]

lgtm

[prasadv, 2015-06-01 18:31:41]

prasadv@chromium.org changed reviewers:
+ prasadv@chromium.org

[prasadv, 2015-06-01 18:31:42]

lgtm

[qyearsley, 2015-06-01 18:35:05]

On 2015/06/01 18:29:29, shatch wrote:
> lgtm

Side comment: I think this CL could have been made as two separate CLs (clean up
symlinks + change checks for empty lists), and that would make it so that each
CL is more specific and has a more descriptive summary line.

[bcwhite, 2015-06-02 00:19:08]

On 2015/06/01 18:35:05, qyearsley wrote:
> On 2015/06/01 18:29:29, shatch wrote:
> > lgtm
> 
> Side comment: I think this CL could have been made as two separate CLs (clean
up
> symlinks + change checks for empty lists), and that would make it so that each
> CL is more specific and has a more descriptive summary line.

I wondered if I should do that.  I'll keep it in mind for next time.

However, it occurred to me after that Python already does the "right thing" when
testing lists with a value of zero (different that a scalar check between None
and zero).  So the test was correct and my change in that file accomplished
nothing.  Odd that it seemed to also fix the problem but perhaps it was an
oddity of the bisect that just never called it.

What I missed was the "not" in the test of the optional parameter.  I just need
to set that to True during the call.

[bcwhite, 2015-06-02 17:22:52]

On 2015/06/02 00:19:08, bcwhite wrote:
> On 2015/06/01 18:35:05, qyearsley wrote:
> > On 2015/06/01 18:29:29, shatch wrote:
> > > lgtm
> > 
> > Side comment: I think this CL could have been made as two separate CLs
(clean
> up
> > symlinks + change checks for empty lists), and that would make it so that
each
> > CL is more specific and has a more descriptive summary line.
> 
> I wondered if I should do that.  I'll keep it in mind for next time.
> 
> However, it occurred to me after that Python already does the "right thing"
when
> testing lists with a value of zero (different that a scalar check between None
> and zero).  So the test was correct and my change in that file accomplished
> nothing.  Odd that it seemed to also fix the problem but perhaps it was an
> oddity of the bisect that just never called it.
> 
> What I missed was the "not" in the test of the optional parameter.  I just
need
> to set that to True during the call.

Still okay with y'all?

[qyearsley, 2015-06-02 17:43:27]

On 2015/06/02 17:22:52, bcwhite wrote:
> On 2015/06/02 00:19:08, bcwhite wrote:
> > On 2015/06/01 18:35:05, qyearsley wrote:
> > > On 2015/06/01 18:29:29, shatch wrote:
> > > > lgtm
> > > 
> > > Side comment: I think this CL could have been made as two separate CLs
> (clean
> > up
> > > symlinks + change checks for empty lists), and that would make it so that
> each
> > > CL is more specific and has a more descriptive summary line.
> > 
> > I wondered if I should do that.  I'll keep it in mind for next time.
> > 
> > However, it occurred to me after that Python already does the "right thing"
> when
> > testing lists with a value of zero (different that a scalar check between
None
> > and zero).  So the test was correct and my change in that file accomplished
> > nothing.  Odd that it seemed to also fix the problem but perhaps it was an
> > oddity of the bisect that just never called it.
> > 
> > What I missed was the "not" in the test of the optional parameter.  I just
> need
> > to set that to True during the call.
> 
> Still okay with y'all?

LGTM! Remember to update the CL description (and maybe title) before submitting
:-)

[bcwhite, 2015-06-02 17:59:01]

> LGTM! Remember to update the CL description (and maybe title) before
submitting
> :-)

Done.

[bcwhite, 2015-06-02 17:59:12]

The CQ bit was checked by bcwhite@chromium.org

[bcwhite, 2015-06-02 17:59:13]

The patchset sent to the CQ was uploaded after l-g-t-m from
simonhatch@chromium.org, prasadv@chromium.org
Link to the patchset: https://codereview.chromium.org/1156963003/#ps20001
(title: "fix for invalid confidence was wrong; now correct")

[commit-bot: I haz the power, 2015-06-02 17:59:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1156963003/20001

[commit-bot: I haz the power, 2015-06-02 18:16:54]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-06-02 18:18:53]

Patchset 2 (id:??) landed as
https://crrev.com/500be8ebfa734283910df8c30de133a219296601
Cr-Commit-Position: refs/heads/master@{#332434}