Adding bibisect repository support to tinbuild2

[Robinson Tryon]

(updating subject line)

On Wed, Jul 18, 2012 at 6:45 PM, Norbert Thiebaud <nthiebaud at gmail.com> wrote:
>>>
>>> + I'd like not to create a new step for that but to fold it in the 'push' step
>>
>> For clarity I'll keep the bulk of the code factored-out into a
>> separate function, but I'll have that function called from within
>> do_push().
>
> Yep, that is fine.

Now that I've done some testing here, I realize that we need some way
to turn off the push-nightlies step while still enabling the creation
of the bibisect repositories. For example, I don't want to bother with
pushing nightlies, especially when I'm testing my commits locally. To
fix I've reordered the day-checking guts of do_push() so that we do
our nightly check first, then call do_bibisect_push(), then (if
pushing nightlies is enabled), call push_nightlies.sh. Both parts are
included in the same tinbuild "phase," but can be enabled
independently.

>> I think that all of the git and file operations should run on most
>> *nix systems, including OSX. There might be a couple of tweaks to make
> ...
> But I'm not so much concerned about the git-foo part of the script but
> rather with what is put in bi-bisect.
> make dev-install does not work on all platform and you do not have a
> */opt/* to 'add' .

Based on the discussion in this thread it sounds like we're pretty
close with an opt/ directory and make dev-install, but we might need
an unpack step for at least OS/Darwin (and maybe for Windows?). To
address these issues (and per suggestion), I've moved the code that
handles finding the opt/ directory, make-install, copying, etc.. into
tinbuild_internals.sh where we can further override it in
platform-specific files such as tinbuild_internals_Darwin.sh.

On Thu, Jul 19, 2012 at 5:11 AM, Norbert Thiebaud <nthiebaud at gmail.com> wrote:
> BTW on mac I though that 'mounting' the dmg and then copying the
> content it over the 'artefact' git should do the trick...
> due to a poor upload bandwidth, I did not bother with it so far since
> I cannot afford to operate such bibisect construction on my box (well
> that is If I want to make it available to others)

The redo-repos.sh script is set up to create two local repos:
'bibisect-repository' and 'fake-remote'. All of the git operations,
even the so-called "remote" operations, happen locally on your box, so
it's perfect for fast testing (or when you don't have much/any upload
bandwidth! :-)

Here's how I create the repos and test the system:

(in the LO checkout)
../buildbot/bin/redo-repos.sh
../buildbot/bin/tinbuild2 -0vx -m debug -p bob

Norbert -- if you have a OSX buildbot available and can help figure
out the darwin-specific pieces, that would be very helpful. I hope
that all we'll have to do is make a darwin version of
copy_build_into_bibisect_repository(). I've just stubbed-in a basic
version, but I really need a mac to test it on -- see the patches for
details, and let me know what works and what needs fixing :-)


Cheers,
--R
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Bibisect-Add-support-for-bibisect-repository-integra.patch.gz
Type: application/x-gzip
Size: 6057 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/libreoffice/attachments/20120720/f17ef21e/attachment-0001.bin>