GSoC 2019 [Buildbot ideas]

[Mojca Miklavec]

On Fri, 29 Mar 2019 at 09:18, Pierre Tardy wrote:
>
> My fear is that this is part of the many stretch goals, and this is beginning to be very optimistic schedule.
> I think it is best to make a great finished GSoC rather than lots of very cool but unfinished mini projects.

I would say that it makes sense to keep all the items you already
wrote (no need to delete them), but move some of them to extension
goals.

Also, there is no point in burning out in the first two weeks by
spending 60 hours per week working on the project, and then quit / be
unable to continue.

It does make sense to play a bit with various pet projects, but we
should really make sure that whatever we really want to achieve is
done in an excellent way, and then other stuff may follow, depending
on time. It's hard to predict how much time certain tasks need, but
it's important to deliver a smaller amount of high quality code rather
than high quantities of unfinished products.

A question to both of you (Pierre, Rajdeep): what's the status of the
boilerplate code you wrote so far? Can any of that be polished,
documented etc., so that it can be officially published?
(I'm really not that familiar with JS.)


Regarding the calendar widget: after something thinking what I believe
would make sense is to pick just a single timestamp and then either
show all the builds before or all the builds after that timestamp (in
decreasing or increasing order, letting the user to browse back and
forth from there; maybe we could even have a button "go one
day/week/month back/forth"). What I miss at the moment is that I might
know that I committed something 3 months ago (I can check the exact
time of commit in repository), but it's non-trivial to find just that
single point in time without manual bisection and lots of
trial-and-error. There is no need to set both start and end time.

Mojca