[Adium-devl] Nightly Builds?

[Chris Forsythe]

I think we talked about finding another mirror to host just the  
nightlies as well, since we're pushing a lot through cachefly at the  
moment.

I agree with David, let's start off small and see how that goes.  
Nightlies bit us pretty bad last time, but you made a very good point  
as to why.

So, this is a yes for me based on waiting for:

1) Another mirror (probably easy, we get offers for this quite a bit.  
Could ask on the blog for another mirror too)

2) Incremental patching within sparkle.

3) I'd like some infrastructure/policies about cutoff points of  
support, things like. Getting organized now instead of organically  
growing the organization later could really help here.

4) Publicizing it in small groups, and expanding that only if the  
need arises and the team agrees to expanding it.

Chris



On Feb 9, 2008, at 6:06 PM, Colin Barrett wrote:

> Tick brought up a good point on IRC: bandwidth costs of this could get
> expensive quickly.  Getting incremental patchingworking will delay
> things a couple days, but don't worry, it's coming.
>
> -Colin
>
> On Feb 9, 2008, at 12:53 PM, Colin Barrett wrote:
>
>> You may have seen some of the commits I've been doing the past few
>> days. In case you haven't, I've been setting up parts of the
>> infrastructure we would need to produce periodic builds automatically
>> (using buildbot) and updating people to newer builds (with Sparkle).
>> Big thanks to my co-conspirator Zac!
>>
>> I'm far enough along now that I think the question of "do we even  
>> want
>> to do this?" needs asking. I (obviously) would argue yes. My main
>> reason is that it gets more people running the current code. We need
>> people testing what we write and filing bugs. The past few releases
>> cycles we've put in a lot of work just fixing regressions once people
>> get alphas and betas -- with nightly builds, we'll have a lot of  
>> those
>> same people looking for problems as we go, so we'd be able to catch
>> regressions when they happen, which hopefully will cut down on
>> debugging time since everything is still fresh.
>>
>> In the past when we did weekly alpha releases (2004), we had problems
>> because we advertised them on the main page of our website[1]. This
>> was clearly not a good idea, and not a mistake we would make again.
>> I'd propose that the URL be announced just in the the topic of  
>> #adium.
>> Maybe put a note up on the beta page telling people to try the
>> nightlies instead, too. Definitely not on the main page, and probably
>> not on the official Adium blog (at least, not right away).
>>
>> I think in addition to announcing too widely, our past bad  
>> experiences
>> with nightly builds were a conflict of culture. We're also a much
>> different project than we were 4 years ago -- back then we hadn't  
>> even
>> released publicly yet, we were still "Adium 2.0", and we used the  
>> blog
>> on the front page to hold conversations back and forth (take a  
>> look!).
>> We're a much more mature and stable project than we were then, I  
>> think
>> we're ready for the responsibility of making sure code on the  
>> trunk is
>> useable by a thousand people.
>>
>> Thoughts? Let's try to arrive at a consensus by Wednesday.
>>
>> -Colin
>>
>> [1]: http://web.archive.org/web/20040121135533/http://www.adiumx.com/
>>
>> _______________________________________________
>> Adium-devl mailing list
>> Adium-devl at adiumx.com
>> http://adiumx.com/mailman/listinfo/adium-devl_adiumx.com
>
>
> _______________________________________________
> Adium-devl mailing list
> Adium-devl at adiumx.com
> http://adiumx.com/mailman/listinfo/adium-devl_adiumx.com