git-try: When -r isn't specified, let the try server decide the revision.

git-try: When -r isn't specified, let the try server decide the revision.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=24848

[Evan Martin, 2009-08-26 01:30:48]

M-A: does this live in depot_tools now?  I'll take down my version if so.

[M-A Ruel, 2009-08-26 01:36:05]

lgtm

[Evan Martin, 2009-08-26 01:43:49]

mattm, you'll have to submit it to the depot_tools site

[mattm, 2009-08-26 01:50:24]

On 2009/08/26 01:43:49, Evan Martin wrote:
> mattm, you'll have to submit it to the depot_tools site

I'll take a look at it.  The depot_tools version is significantly different, so
it may not apply.

[Evan Martin, 2009-08-26 02:38:10]

On 2009/08/26 01:50:24, mattm wrote:
> On 2009/08/26 01:43:49, Evan Martin wrote:
> > mattm, you'll have to submit it to the depot_tools site
> 
> I'll take a look at it.  The depot_tools version is significantly different,
so
> it may not apply.

I think you can just clobber the depot_tools one with the one in my tree.  (it
may just be an older version)

[chase, 2009-08-26 04:18:50]

On 2009/08/26 02:38:10, Evan Martin wrote:
> I think you can just clobber the depot_tools
> one with the one in my tree.  (it may just be
> an older version)

The depot_tools one is newer than the one in your tree, Evan.  Let's not clobber
it. :)

Matt, remove lines 176-177 in the latest version of depot_tools/git-try, then
remove the GetRevision method.  That should do what you want.

[mattm, 2009-08-26 21:53:02]

On 2009/08/26 04:18:50, chase wrote:
> On 2009/08/26 02:38:10, Evan Martin wrote:
> > I think you can just clobber the depot_tools
> > one with the one in my tree.  (it may just be
> > an older version)
> 
> The depot_tools one is newer than the one in your tree, Evan.  Let's not
clobber
> it. :)
> 
> Matt, remove lines 176-177 in the latest version of depot_tools/git-try, then
> remove the GetRevision method.  That should do what you want.

Well, I made the change.  I don't actually know that it's what I want ;)

(On the one hand, I think this makes git-try behave like gcl try does.  On the
other hand, using the same revision as your checkout may actually more useful
when lkgr is frequently out of date?)

[chase, 2009-08-29 00:29:08]

> Well, I made the change.  I don't actually know that it's what I want ;)
> 
> (On the one hand, I think this makes git-try behave like gcl try does.  On the
> other hand, using the same revision as your checkout may actually more useful
> when lkgr is frequently out of date?)

Code change lgtm.  I assumed git-try already acted by default like how your
patch changes it to act.  It seems less surprising to act like gcl and use lkgr
by default.  M-A / Evan will speak up if they prefer one way over the other.

[M-A Ruel, 2009-08-29 01:09:47]

On 2009/08/29 00:29:08, chase wrote:
> > Well, I made the change.  I don't actually know that it's what I want ;)
> > 
> > (On the one hand, I think this makes git-try behave like gcl try does.  On
the
> > other hand, using the same revision as your checkout may actually more
useful
> > when lkgr is frequently out of date?)
> 
> Code change lgtm.  I assumed git-try already acted by default like how your
> patch changes it to act.  It seems less surprising to act like gcl and use
lkgr
> by default.  M-A / Evan will speak up if they prefer one way over the other.

I prefer gcl-like behavior. In the last your, I got 2 reports of try failures
and I highly suspect they were using git-try while HEAD was broken.

M-A

[mattm, 2009-08-29 02:01:30]

On 2009/08/29 01:09:47, Marc-Antoine Ruel wrote:
> On 2009/08/29 00:29:08, chase wrote:
> > > Well, I made the change.  I don't actually know that it's what I want ;)
> > > 
> > > (On the one hand, I think this makes git-try behave like gcl try does.  On
> the
> > > other hand, using the same revision as your checkout may actually more
> useful
> > > when lkgr is frequently out of date?)
> > 
> > Code change lgtm.  I assumed git-try already acted by default like how your
> > patch changes it to act.  It seems less surprising to act like gcl and use
> lkgr
> > by default.  M-A / Evan will speak up if they prefer one way over the other.
> 
> I prefer gcl-like behavior. In the last your, I got 2 reports of try failures
> and I highly suspect they were using git-try while HEAD was broken.
> 
> M-A

Okay, I committed it.