AOSP contributions: looking back, looking ahead

[Jean-Baptiste Queru]

February is over, which was our first full month with Gerrit running
again after our 2012 outage.

Approximately 120 contributions were merged in February. The data
shows that the pace at the end of the month was much better than at
the beginning, which shows that we've getting back to speed.

Also, we've just fixed an issue with Gerrit, such that I don't need to
personally serialize submissions of reviewed changes or synchronize
them against other processes. That will save me hours of work.

Still in the "good news" bucket, Conley has been making progress
getting our automated verification process going again. It's not fully
functional yet, but it's already providing valuable data.

Note however that I'll have to temporarily step away from my Android
tasks for a good chunk of time this month, and AOSP activity will
probably be slower during that period. I'll be back. No need to be
worried when you can simply be patient.

JBQ

--
Jean-Baptiste M. "JBQ" Queru
Software Engineer, Android Open-Source Project, Google.

Questions sent directly to me that have no reason for being private
will likely get ignored or forwarded to a public forum with no further
warning.

[Kenneth Crudup]

On Mar 1, 4:14 pm, Jean-Baptiste Queru <j...@android.com> wrote:

> ... and AOSP activity will probably be slower during that period. ... No need to be worried when you can simply be patient.

Too bad for that- seeing new merges pop up during "repo forall git
fetch -all" was like getting new AOSP presents every day :) (It's just
as well; now maybe I'll have an uptime > 1 day on my Xoom!)

[Jean-Baptiste Queru]

Well, we fixed the Gerrit issue where I was the only person who could
submit changes, and now everyone in the Android team can do so (and
they already do, I'm seeing that 6 different engineers submitted
changes since I opened up the floodgates).

JBQ

--

[enh]

am i the only one who's having trouble with .netrc? i find have to cat it and copy & paste my gerrit username and password, and i'm pretty sure it's supposed to be a bit more automatic than that :-)

other than that, working in AOSP does seem to be back to normal. thanks!

--
Elliott Hughes - http://who/enh - http://jessies.org/~enh/
NIO or JNI questions? Mail me/drop by/add me as a reviewer.

[Jean-Baptiste Queru]

Somewhere on my whiteboard is a note that I need to document this cleanly.

There are some issues with teh .netrc setup around machine names.

The machine name for the downloads is android.googlesource.com
The machine name for the uploads is android-review.googlesource.com

Sometimes the password generator provides a sample for
www.googlesource.com, which works for neither.

The same username/password works for all of googlesource.com, but the
syntax of .netrc forces to specify each hostname individually.

JBQ

[enh]

that makes perfect sense; my .netrc has the download hostname. i'll add a line for the upload hostname too. thanks!

[Gergely Kis]

Hi,

2012/3/2 Jean-Baptiste Queru <j...@android.com>:
[...]

>
> Note however that I'll have to temporarily step away from my Android
> tasks for a good chunk of time this month, and AOSP activity will
> probably be slower during that period. I'll be back. No need to be
> worried when you can simply be patient.
>
>

This is definitely bad news for us. We submitted our DTMF telephony
contribution on the 17th, and there is still no review activity on it,
not even a verified flag.
For reference, here are the 3 changes:
https://android-review.googlesource.com/#/c/32821/
https://android-review.googlesource.com/#/c/32820/
https://android-review.googlesource.com/#/c/32810/

The problem is not that there is a backlog, but that there are
submissions which seem to be falling through the cracks and never get
reviewed (and for us, that our submission is one of these :) ), while
a lot of others are not only reviewed but merged as well. It is
certainly possible that we did something wrong during submission, but
if we don't get any feedback, we won't be able to fix these issues.

Best Regards,
Gergely

[Jean-Baptiste Queru]

The review delays are definitely an issue.

In order to avoid having contributions fall through the cracks, I
explicitly try to look at the contributions that have been inactive
the longest. That way, every contribution gets looked at...
eventually. Ignoring changes from the old Gerrit, the line is
currently at Jan 19, i.e. all the contributions from before Jan 19
have been looked at at least once.

Verifying the changes is actually the biggest bottleneck at this
point. Doing so takes time. Even if I can dedicate an entire day to
verifying changes, I can still only look at a few dozen changes a day,
and I have so many tasks that prevent me from verifying changes all
day.

JBQ

--

[Gergely Kis]

Hi,

2012/3/5 Jean-Baptiste Queru <j...@android.com>:

> The review delays are definitely an issue.
>
> In order to avoid having contributions fall through the cracks, I
> explicitly try to look at the contributions that have been inactive
> the longest. That way, every contribution gets looked at...
> eventually. Ignoring changes from the old Gerrit, the line is
> currently at Jan 19, i.e. all the contributions from before Jan 19
> have been looked at at least once.

Basically this means for me that there is about zero chance to get our
contribution into master (and hopefully into the next update of the
platform), unless we can find someone besides you who can push it
through the verification / review cycle.

The question is: can you help us find this person?

I already added everyone as reviewer who we talked to in earlier
iterations of this patch, but there has been no response until now. It
would also help if we could just get a response like "we won't have
time for this before the next release" or "we are working on
refactoring the whole telephony stack, so your patch would not apply
anymore to the internal tree".

Thanks,
Gergely