svn commit: trunk/busybox: scripts

Bernhard Fischer rep.nop at aon.at
Wed May 10 11:45:20 PDT 2006


On Wed, May 10, 2006 at 01:53:09PM -0400, Rob Landley wrote:
>On Tuesday 09 May 2006 7:51 pm, Mike Frysinger wrote:
>> On Tuesday 09 May 2006 10:24, Rob Landley wrote:
>> > On Tuesday 09 May 2006 7:34 am, Bernhard Fischer wrote:
>> > > You, especially as the official maintainer (but just as the rest of us)
>> > > are supposed to check if the stuff you intend to checkin actually
>> > > works. This is supposed to be done in a *separate* tree where only the
>> > > hunks you intend to checkin are applied to a tree which is otherwise
>> > > pristine..
>> >
>> > I'm confused.  For the -stable branch you insist I use svn.  For -devel,
>> > you want me to _stop_ using it?
>>
>> multiple copies of the same repo checked out ... not that hard, i usually
>> keep 'busybox' and like 'busybox-clean' side by side
>
>I made a script that anonymously checks out the repository and builds it 
>(allnoconfig, allbareconfig, and defconfig with a distclean between each) 
>out-of-tree.  Yesterday I added a call to "make help".
>
>But that only tests what I just checked in (so I can fixup the repository in 
>the next checkin; I build and test in my tree before checking in but that 
>doesn't mean the checkin gets everything I needed to), and it doesn't test 
>everything.
>
>I should make it run the test suite and show me the failures...

You misunderstood what i said.

To avoid to put stuff into the repository that you do not want to be
checked in and to check if a given changeset is doing what it is
supposed to do without relying on any other lingering changes, cp a
clean checkout to a scratch-dir, apply the changeset you intend to put
into the repo, build, test, and then check in this changeset out of the
*scratch* copy.

No chance that unwanted stuff ends up being checked in if you use a
*separate* cp which *only* contains the one changeset you want to put in.




More information about the busybox mailing list