CI Requirements - Lessons Not Learnt?

[Ben Cooksley]

On Fri, Jan 6, 2017 at 12:38 AM, Martin Gräßlin <mgraesslin at kde.org> wrote:
> Am 2017-01-05 11:20, schrieb Ben Cooksley:
>>
>> On Thu, Jan 5, 2017 at 10:28 PM, Martin Gräßlin
>> <privat at martin-graesslin.com> wrote:
>>>
>>> Am 2017-01-05 09:44, schrieb Ben Cooksley:
>>>>
>>>>
>>>> Hi all,
>>>>
>>>> It seems that my previous vocal complaints about system level /
>>>> serious impact dependency bumps on the CI system have gone completely
>>>> unnoticed by (some) members of our Community.
>>>>
>>>> This was demonstrated earlier this week when components of Plasma
>>>> bumped their version requirements for XKBCommon and Appstream-Qt -
>>>> without even a thought about notifying Sysadmin or checking which
>>>> version the CI had, until their builds broke.
>>>>
>>>> Neither of these is easy to fix at this stage, as the system base is
>>>> now too old to receive updates such as these. Base upgrades require a
>>>> full rebuild of everything on the CI system, and usually involve
>>>> significant additional churn and is a process that must be done
>>>> roughly twice a year, depending on dependency bump demands.
>>>>
>>>> Does anyone have any suggestions as to how we may avoid this in the
>>>> future?
>>>
>>>
>>>
>>> I have a few questions here:
>>>
>>> 1) Where is this requirement to check with sysadmins codified? So far I
>>> was
>>> only aware of dependency freeze.
>>
>>
>> It's been codified since the PIM Qt 5.6 / WebEngine debacle, where
>> Sysadmin had to rush delivery of Qt 5.6 to the CI system because the
>> whole of PIM broke when they started using QtWebEngine. That was
>> around March/April 2016, my mail archives can't seem to find the exact
>> thread though.
>
>
> I'm sorry Ben, but I fear "sending out a mail about an issue with PIM"
> doesn't
> qualify as codifying it. Given what we have it looks like this did not reach
> the
> target audience. And neither will this thread.

Uhm, it was far more than a single email. It was quite the thread, and
if memory serves was on at least one of the mailing lists this thread
was on.
One of the key things that came out of that thread was to ask for
things to be bumped well in advance if a newer version is needed.

I'm disappointed that you think that email threads have insufficient
reach given it's one of our communities principal means of
communication.

>
> This needs to change the process, the way KDE develops software. It needs to
> be
> listed in the release schedule (is not, I checked), maybe reviews need to be
> acked by release managers, etc.

It seems to be part of the process for many others already, so not
sure what needs to change. The gpgme transition went quite well for
PIM, and other applications developers have reached out and asked
about version upgrades to various libraries which were in their area
of interest which we have sorted out easily enough.

>
>>
>>> 2) How can we easily check what build.kde.org has? Looking at cmake
>>> output
>>> is not a sufficient way as it gives me wrong information
>>
>>
>> If CMake is outputting wrong information, then your CMakeLists.txt
>> can't make the appropriate decisions as to whether the available
>> version is suitable, so i'd say you've got bigger problems here that
>> need to be addressed first.
>
>
> Cmake feature summary says: "required version >= 0.5" and that's for all
> found
> depeendencies. Unfortunately no information at all in the feature summary
> about
> the actual version.

Quoting the KWin CMake log...

15:08:41 -- Found Wayland_Client:
/usr/lib/x86_64-linux-gnu/libwayland-client.so (found version
"1.12.90")
15:08:41 -- Found Wayland_Cursor:
/usr/lib/x86_64-linux-gnu/libwayland-cursor.so (found version
"1.12.90")
15:08:42 -- Found Wayland_Egl:
/usr/lib/x86_64-linux-gnu/libwayland-egl.so (found version "11.0.2")
15:08:42 -- Found Wayland:
/usr/lib/x86_64-linux-gnu/libwayland-client.so;/usr/lib/x86_64-linux-gnu/libwayland-cursor.so;/usr/lib/x86_64-linux-gnu/libwayland-egl.so
(found suitable version "1.12.90", minimum required is "1.2") found
components:  Cursor Egl
15:08:42 -- Could NOT find XKB: Found unsuitable version "0.5.0", but
required is at least "0.7.0" (found
/usr/lib/x86_64-linux-gnu/libxkbcommon.so)
15:08:42 -- Found Libinput: /usr/lib/x86_64-linux-gnu/libinput.so
(found suitable version "1.5.2", minimum required is "1.5")
15:08:42 -- Found UDev: /usr/include
15:08:42 -- Found Libdrm: /usr/lib/x86_64-linux-gnu/libdrm.so (found
suitable version "2.4.64", minimum required is "2.4.62")
15:08:42 -- Found gbm: /usr/lib/x86_64-linux-gnu/libgbm.so (found
version "11.0.2")

The summary may not be of any use, but the rest of the output has the
information you're after.

>
>>
>> In any case, you can see the Docker log of the container being
>> generated at https://build.kde.org/job/create_ubuntu_slave-ange/
>
>
> and where do I find this information if I would not have asked in a mail?
> This is very related to properly codifying and documenting such
> requirements.

Pretty sure i've pointed it out to you (among others) on IRC more than once.
It's true that there isn't written documentation for the CI system as
such, but it's a highly organic system subject to regular change - so
documentation such as that can easily get outdated.

>
> You cannot tell people "don't introduce new dependencies", without telling
> them
> how to check.
>
>>
>>> 3) What should we do when build.kde.org does not have the requirement?
>>
>>
>> You have to file a Sysadmin ticket, also tagging the project
>> 'build.kde.org' at the same time.
>
>
> And then? What's the process then? How long do we have to expect this to go?
> Would it allow to block a finished feature or an important bug fix? Would we
> be
> forced to write ifdef hackery?
>
> Sorry, but I'm not thrilled by this process.
>
> What matters to me is getting out good software to our users. And for that I
> have
> a hard requirement I have to hit: dependency freeze.

The process would depend on what is needed.

If it is something we simply don't have installed, or a suitable
version is available in backports or another usable repository then
it's fairly trivial and the turnaround would probably be a day or two.
That's something you can accelerate by providing the names of the
needed packages for our base system(s), along with details of PPAs,
etc which may be needed. If it's something like Qt on the other hand,
that is far more involved and could easily span to a few weeks.

Note that you should probably be providing notice to the community at
large who also build your software anyway, so that developers can get
their systems ready for the dependency bump. Otherwise you're breaking
their workflows as well. By all accounts you broke the builds of other
Plasma developers, which indicates to me that this dependency bump
wasn't announced very well at all.

>
>>
>>>
>>> It should be rather obvious that we don't introduce new dependencies
>>> because
>>> we like to. There is a very important software reason to it.
>>> That's the case for the xkbcommon dependency increase. Should I have let
>>> the
>>> code broken as it was, expecting half a year of bug reports till
>>> build.kde.org has the base upgraded to Ubuntu 16.04?
>>
>>
>> That's what #ifdef is for...
>
>
> I see you volunteer to:
> 1. write the ifdef
> 2. adjust the unit test to skip
> 3. Inform distros that the reported minimum version is wrong, that in truth
> it
>   requires a newer version than reported
> 4. handle all the bug reports related to it
>
> If not, please don't suggest ifdef. We all know that it comes with a huge
> cost.
> A cost I decided is too high in this case. After all I had many people
> complain
> about it and you can imagine how annoyed I am about the broken build.
>
> If it were as simple as an ifdef, we would have done it, wouldn't we?
>
>>
>>>
>>> If I have to degrade the quality of the product for serving the CI, I and
>>> all users have a problem. And this is currently the only alternative. The
>>> quality of our product is highly at risk as our changes are no longer
>>> compile tested. This is a huge problem for the release of Plasma 5.9. On
>>> the
>>> other hand I cannot revert the dependency change as that would break
>>> tests
>>> or introduce the broken code again. So actually we are caught between a
>>> hard
>>> and a rock place.
>>>
>>> When I increased the dependency I had the dependency freeze of Plasma 5.9
>>> in
>>> mind. That's the one target I have to hit from release process currently.
>>> Also I had to consider a social aspect here. I asked xkbcommon devs to do
>>> the release. I would have feeled ashamed if we asked for the release and
>>> then don't use it. For me it was from a social point of view a very high
>>> requirement to ship with the dependency in the next release after
>>> xkbcommon
>>> release.
>>>
>>> If we have to wait an arbitrary time till build.kde.org has upgraded the
>>> base, maybe the choice of the base is not sufficient. E.g. I asked
>>> openSUSE
>>> about this dependency weeks ago. Actually a few days after xkbcommon had
>>> the
>>> release and it was already shipped in tumbleweed. Similar for Mesa 13
>>> which
>>> I'm also eagerly waiting for build.kde.org to fetch it.
>>
>>
>> Mesa 13 is news to me.
>
>
> Oh we talked about it months ago when I told you that Mesa 13 will have a
> new way
> to get a surfaceless context. That was prior to the release of Mesa 13,
> though.

Fair enough.
I've yet to see any request for it to be made available, which means
it's hard to plan for.

>
> Cheers
> Martin

Regards,
Ben