On Wed, Sep 2, 2009 at 16:00, Yves Goergen<nospam.list_at_unclassified.de> wrote:
> On 02.09.2009 16:57 CE(S)T, David Weintraub wrote:
>> That's a shame really because you would find CruiseControl.NET to be
>> extemely useful in your development.
> Hm, from those two entry pages, CCNet doesn't look too useful to our
> project. We don't need a build server. And we do commit every few hours
> already. And we can talk to each other if anyone wants to do big
Don't dismiss it unless you've tried it.
>> You should not be "pushing the build button" for official builds. You
>> should not be doing any manual steps in order to produce your builds.
>> That's not only error prone, but makes the build process totally
>> dependent upon you. Official builds should take place on a server
>> specifically designed for that purpose and should be 100% automated.
> Everyone who works on the project (1 to 3 people) can build the
> application and publish the binary. There's no need for a highly
> sophisticated build server environment. Testing was not paid by the
> client so we don't do it a lot other then what a developer usually does.
Then how do you know your build process is consistently & correctly followed?
If a developer breaks something, wouldn't you rather know at 9 AM when
he does the commit than at 5 PM when the next developer attempts to do
Why are you spending valuable developer time producing & testing
builds when you could have a system doing it for you automatically?
As for the "it compiles, ship it!" methodology... Formal Testing
should be part of every development process, especially when it's
under contract. Surely there's some measure of quality specified in
the contracts with your clients. I do exclusively internal development
and we don't release anything for acceptance testing until someone on
the development staff has successfully executed the full test plan.
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
Received on 2009-09-02 22:13:18 CEST