[Swift-devel] Coaster capabilities for release 0.9

[Michael Wilde]

Raising the urgency of testing coasters for 0.9 is inconsistent with 
saying there is "little point in putting coaster changes into a release".

If we are willing to insert more point releases between 0.9 and 1.0, or 
we dont attach special significance to 1.0, I am happy to let 0.9 go. 
Otherwise, I want to delay 0.9. Thats a separate but related question, 
and good to decide before we call the next release 0.9.

That aside, I want to set a firm goal that a coaster feature we can call 
production-quality be in the next point release after the one currently 
in candidate state. Its OK to have known limitations, but it should do a 
useful core set of things very well, with tests to validate it, and 
documented so users know what they can and can not expect it to do. (Ie 
when to use it and when not).

I can see that this may take 8 weeks or more. I still want a list of the 
known coaster improvements needed, and a time estimate for doing them, 
and design discussion on any features that need it (such as coasters 
that work well on OSG). Otherwise we cant tell if coasters needs 8 weeks 
or 80 weeks. I started such a list in prior emails.

Mihael, please merge that with a list of issues you see from emailed 
reports, other issues you know of or have concerns about, and send out 
to swift-devel.

Now that we have some experience with coasters - much of it very 
positive - we should bear down and focus on making it production 
quality. I hear the dismissal of my previous statement on that as a 
mom-and-apple-pie rant, but a list of know deficiencies and how you plan 
to address them is the concrete part of what Im asking for when I say this.

- Mike



On 4/21/09 11:28 AM, Ben Clifford wrote:
> On Tue, 21 Apr 2009, Michael Wilde wrote:
> 
>> Which of these can make it into 0.9? (I would like see 0.9 have a highly 
>> usable coaster release with a converging set of capabilities. I would 
>> argue to delay 0.9 by up to 2 weeks to get it in, if needed enhancements 
>> are close)
> 
> A 0.9 release candidate has already been made. Bugs would have to be 
> fairly serious in order to prevent me releasing it.
> 
> Coasters, as an in-development feature, not a production-quality feature, 
> pretty much shouldn't get to stop or delay releases unless they are 
> substantially worse than swift 0.8 coasters.
> 
> Post 0.9rc2 changes will end up in the next swift release, which should be 
> roughly 8 weeks after the 0.9 release. That is not so long to wait.
> 
> If there is likelihood that the advice from now onwards until that next 
> release will be "download the latest svn" to coaster users, then there is 
> little point in putting coaster changes into a release.
>