Fixing #654506 and #674386 in Wheezy

[Felipe Sateler]

On Wed, Jul 11, 2012 at 10:30 AM, Mehdi Dogguy <mehdi at dogguy.org> wrote:
> On 11/07/12 16:01, Felipe Sateler wrote:
>>
>> On Wed, Jul 11, 2012 at 8:20 AM, Mehdi Dogguy<mehdi at dogguy.org>
>> wrote:
>>>
>>> Hi,
>>>
>>> We would like to fix #654506 and #674386 in Wheezy. Unfortunately,
>>> we are not able to accept supercollider/1:3.5.2-1 from Unstable
>>> since the changes are quite large.
>>
>>
>> I think you mean 1:3.5.3~repack-1?
>
>
> Yes, sorry. It was a bad copy/paste :/
>
>
>> That is what's currently in unstable, and 1:3.5.2-1 was uploaded
>> before the freeze. Unfortunately, it couldn't migrate because it
>> failed to build on non-x86 archs. We are currently working on fixing
>> that. So, in a way, the changes are not that large ;).
>>
>
> We don't seem to have the same definition of "large". For this specific
> case, the changes between the unblocked version and sid's current
> version look like:
>
> $ debdiff supercollider_3.5.2-1.dsc supercollider_3.5.3~repack-1.dsc \
>   | diffstat | tail -n1
>  3040 files changed, 5266 insertions(+), 581639 deletions(-)
>
> This pretty looks as "large". Ignoring the bits that were deleted when
> repacking, the debian/ directory, etc… this leads us to:
>
>  53 files changed, 746 insertions(+), 701 deletions(-)
>
> which is nicer indeed but still qualifies as large.

I made some local git branches with the upstream source of 3.5.2 and
3.5.3, with patches applied.

Updating to 3.5.3 allowed us to drop all the 3.5.2 patches:
$ git show --stat  3.5.2-withpatches'^' | tail -1
 7 files changed, 135 insertions(+), 100 deletions(-)

So, taking into account this, the stat becomes:

$ git diff 3.5.2-withpatches..3.5.3-withpatches --stat \
   | tail -1
 52 files changed, 631 insertions(+), 198 deletions(-)


However, a big chunk of that is documentation updates:

$ git diff 3.5.2-withpatches..3.5.3-withpatches --stat \
   -- HelpSource/| tail -1
 18 files changed, 439 insertions(+), 131 deletions(-)

That leaves as with a diff of:
 34 files changed, 192 insertions(+), 67 deletions(-)

Of that, most of it is bugfixes, and an un-deprecation of a few methods.


>
> Why did you import 3.5.3 instead of working on fixing 3.5.2? (I'm not
> sure it is relevant now but that might help us to understand the
> situation better).

Mostly because it allowed us to drop the patches we had. Also,
upstreams release management seems sane enough, commits on the 3.5
branch are mostly cherry-picked from the master branch plus
documentation fixes.


>
>
>> I had planned to mail d-r after we got the last round of fixes ready.
>> Is there a chance we can convince you to let 3.5.3 migrate to
>> testing?
>>
>
> We would prefer targeted fixes based on the version of testing.

I understand. But on the other hand, we would prefer shipping
upstreams latest version, which is why I asked if there was a chance
we could convince you.
In particular, since 3.5 sc has a new Qt based widget system, and
debian does not have any other sc widget system (AFAICT, they were all
third party), wheezy users would not be able to build SC GUIs.
Dan can probably tell of more advantages of 3.5 over 3.4.
That's why I asked if there was a chance that we could convince you. I
wasn't asking if we had clearance yet.


-- 

Saludos,
Felipe Sateler