Skip to Content.
Sympa Menu

sm-discuss - Re: [SM-Discuss] Spell quality checking

sm-discuss AT lists.ibiblio.org

Subject: Public SourceMage Discussion List

List archive

Chronological Thread  
  • From: David Kowis <dkowis AT shlrm.org>
  • To: SM-Discuss <sm-discuss AT lists.ibiblio.org>
  • Subject: Re: [SM-Discuss] Spell quality checking
  • Date: Thu, 01 Oct 2009 13:15:35 -0500

flux wrote:
> Andraž 'ruskie' Levstik (ruskie AT codemages.net) wrote [09.10.01 08:34]:
>>> I'm all for having proper unit tests myself, but let's not get too
>>> carried away with where we apply our efforts. A unit test that ensures a
>>> config query actually results in applying the user's choice is something
>>> we can do. A unit test that guarantees there are no memory leaks or
>>> segfaulting operations in a final product is not something that should
>>> be our goal as a distro.
>> You're thinking in the wrong direction. We don't care about specific
>> memleaks/segfaults etc... We just care that the app should do what it
>> says on the tin.
>
> That is exactly what I'm talking about. The memleaks and segfaults were
> just examples. Making sure a program does what it claims is the
> responsibility of those writing the program. Our unit tests should
> ensure that when a user is asked if they want feature X and the user
> answers no, that feature X had better be disabled in the spell being
> built. That's our responsibility.
>
> Making all the programs work together is also something that should
> happen, but in my opinion not to the extent that we should be devoting
> our time to writing unit tests which ensure precisely that. Upstream
> should be doing this. We are the end-users of those build products, not
> the developers of them. The users of our distro are the end-users of the
> built (packaged) product, which we develop.
>

And we're writing tests to ensure that our build/package of the project
with the configure options we passed to it, actually built it the way we
configured it to. When we pass --with-foobar, we determine that
libfoobar is actually linked to the binary. Proving that our build
system does indeed build the software the right way.

If we wanted to extend it beyond that to proving that the webbrowser
works the way it should, we could do that, but that's not the initial point.

David


Attachment: signature.asc
Description: OpenPGP digital signature




Archive powered by MHonArc 2.6.24.

Top of Page