SDK constraint and README for barback.

SDK constraint and README for barback.

BUG=https://code.google.com/p/dart/issues/detail?id=14862
R=jmesserly@google.com, nweiz@google.com

Committed: https://code.google.com/p/dart/source/detail?r=30200

[Bob Nystrom, 2013-11-06 23:02:26]

CC'd folks, this is to try to handle the fact that barback is currently tightly
coupled to the specific version of the SDK it's running on. It means (until
barback itself is stable), we'll have to publish a new version of barback when
we publish new SDK builds.

What do you think? Any other ideas?

- bob

[Jennifer Messerly, 2013-11-06 23:08:12]

if this is the solution ... I think we should revert some of my changes
yesterday to publish_pkg.py. In particular, we must ensure *every* SDK releases
with a compatible "barback", right? Otherwise users of that SDK are hosed...

our old script handled this: it stamped a version and SDK constraint
automatically. perhaps we should restore the code but it only kicks in for
barback.

honestly--if barback is this tightly coupled, shouldn't it ship with the SDK?

[nweiz, 2013-11-07 00:30:47]

lgtm

lgtm

[Bob Nystrom, 2013-11-07 00:47:50]

Good point. Added a script to automatically publish this when we cut an SDK
build.

PTAL!

[Jennifer Messerly, 2013-11-07 00:49:47]

lgtm

[nweiz, 2013-11-07 00:55:55]

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py
File tools/publish_barback.py (right):

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:15: # with the user's SDK.
Should we mention here that this process will change when the wire protocol
becomes more stable?

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:53: # decided.
Is there a bug number you can list here?

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:73: constraint = '>=%d.%d.%d <=1.0.0' % (major, minor,
build)
I think this should be ">=%d.%d.%d <1.1.0", in case 1.0.0 gets hotfixed.

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:83: print 'Publishing version %s of barback to pub with
SDK constraint "%s".' % (version, constraint)
Long line

[Bob Nystrom, 2013-11-09 01:24:26]

Dan, do you have any thoughts on if you want me to land this or not?

- bob

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py
File tools/publish_barback.py (right):

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:15: # with the user's SDK.
On 2013/11/07 00:55:55, nweiz wrote:
> Should we mention here that this process will change when the wire protocol
> becomes more stable?

It's actually the barback public API, not the wire protocol. Pub has both sides
of the wire protocol. It just implements the client side by implementing
barback's public API.

But done.

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:53: # decided.
On 2013/11/07 00:55:55, nweiz wrote:
> Is there a bug number you can list here?

Nope. I don't know if it makes sense to file a bug or not.

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:73: constraint = '>=%d.%d.%d <=1.0.0' % (major, minor,
build)
On 2013/11/07 00:55:55, nweiz wrote:
> I think this should be ">=%d.%d.%d <1.1.0", in case 1.0.0 gets hotfixed.

Done.

https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
tools/publish_barback.py:83: print 'Publishing version %s of barback to pub with
SDK constraint "%s".' % (version, constraint)
On 2013/11/07 00:55:55, nweiz wrote:
> Long line

Done.

[dgrove, 2013-11-12 09:56:49]

On 2013/11/09 01:24:26, Bob Nystrom wrote:
> Dan, do you have any thoughts on if you want me to land this or not?
> 
> - bob
> 
> https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py
> File tools/publish_barback.py (right):
> 
>
https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
> tools/publish_barback.py:15: # with the user's SDK.
> On 2013/11/07 00:55:55, nweiz wrote:
> > Should we mention here that this process will change when the wire protocol
> > becomes more stable?
> 
> It's actually the barback public API, not the wire protocol. Pub has both
sides
> of the wire protocol. It just implements the client side by implementing
> barback's public API.
> 
> But done.
> 
>
https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
> tools/publish_barback.py:53: # decided.
> On 2013/11/07 00:55:55, nweiz wrote:
> > Is there a bug number you can list here?
> 
> Nope. I don't know if it makes sense to file a bug or not.
> 
>
https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
> tools/publish_barback.py:73: constraint = '>=%d.%d.%d <=1.0.0' % (major,
minor,
> build)
> On 2013/11/07 00:55:55, nweiz wrote:
> > I think this should be ">=%d.%d.%d <1.1.0", in case 1.0.0 gets hotfixed.
> 
> Done.
> 
>
https://codereview.chromium.org/63363002/diff/50001/tools/publish_barback.py#...
> tools/publish_barback.py:83: print 'Publishing version %s of barback to pub
with
> SDK constraint "%s".' % (version, constraint)
> On 2013/11/07 00:55:55, nweiz wrote:
> > Long line
> 
> Done.

LGTM.

[Bob Nystrom, 2013-11-12 18:57:25]

Committed patchset #4 manually as r30200 (presubmit successful).