Re: Ports & Packages infrastructure & feature import. (fwd)

[Marc Balmer <mbalmer%NetBSD.org@localhost>]

Am 03.09.11 19:39, schrieb Loganaden Velvindron:
> Wouldn't it be easier to rely on experienced developers in
> that area to review your diff ?
> 
> I already saw some commit messages like that.
> 
> ATF's use is becoming mandatory. Can't the same be done
> with systematic peer review of non-trivial changes ?

real man don't top-post.

> 
> 2011/9/3 Marko Schütz Schmuck <MarkoSchuetz%web.de@localhost>:
>> At Sat, 3 Sep 2011 09:09:31 +0300,
>> Jukka Ruohonen wrote:
>>>
>>> On Fri, Sep 02, 2011 at 10:15:49AM -0400, Marko Schütz Schmuck wrote:
>>>> It might also be a good idea to automate integration testing before
>>>> committing and reject failed integrations. There is an IMNSHO
>>>> excellent tool in Aegis (aegis.sf.net).
>>>
>>> I agree that (automated) "continuous integration" is a good idea. However,
>>> I doubt whether it scales to projects with several millions of lines of 
>>> code.
>>
>> There are relatively few lines that change with each change set and
>> there are few targets that will be impacted. The files generated from
>> a build prior to integration of the change are kept in the project
>> directory. The build tool needs to be capable of using a path (VPATH)
>> to search for files or alternatively the build directory can be
>> populated with links to the changeset's baseline. Also, not all tests
>> are run automatically for each changeset: the developer specifies and
>> submits the tests that the changeset must satisfy (and that the
>> previous version must fail). Of course, regression tests are also
>> possible and aegis can suggest regression tests based on the files in
>> the changeset and the tests that were run for these files in the past.
>>
>> Best ragards,
>>
>> Marko
>>
> 
> 
> 


-- 
  \~~~~~.                The NetBSD Foundation
   \~~~~~'               Marc Balmer, Developer / Marketing
  NetBSD
     \                   mbalmer%NetBSD.org@localhost   http://www.NetBSD.org/