Gsoc 18 Project | Collect build statistics

[Mojca Miklavec]

Dear Vishnu,

I have a bunch of suggestions written below. Community bonding should
usually be used to getting familiar with our codebase, our tools, our
communication channels, getting to know people, ...

Since you are creating a standalone product, this community bonding
could be slightly different (not so much need to learn our base code
by heart), but it should warm you up and get you up to full speed by
the time the coding period starts.

The most important parts are:
- getting MacPorts installed somewhere where you have regular access
- creating a repository and finish/improve the planning, I would like
to see the database design finalized, finish the sample HTML document,
...
- creating tickets with milestones

I hope others will be able to comment on my suggestions (they are not
universal truths :)



INSTALLING MACPORTS, BUILDBOT

(a.1) We explained that owning a Mac was not a requirement for working
on this project, but it would be orders of magnitude easier to work on
some tasks if you had a regular access to MacPorts. This could either
be:
  - SSH access to a remote Mac or VM
  - A Linux machine (even if Raspberry PI) or a VM with Linux

You should start working on this as soon as possible because it would
be a pity to waste precious time during the coding period. MacPorts
can theoretically be installed on a UNIX machine, but there might be
problems of one kind or the other. We can help you circumvent the
problems, but you need to star working on this ASAP.

Please let us know what your plan is.

(a.2) With MacPorts you should be able to install mpstats and use it
to submit statistics (to either the official server or your new server
once ready).

(a.3) It would probably be quite helpful if you manage to install
buildbot and replicate our setup (at least to some extent). But (a.1)
is a strong requirement for that, else it makes no sense.


SQL & OTHER TOOLS

(b.1) This is an absolute must:
    https://www.coursera.org/learn/sql-for-data-science#
along with some further understanding of primary keys, indices,
foreign keys, ...

Let us know whether you managed to subscribe and how you are
progressing, whether you have any further questions etc.

(b.2) I found two (fully optional) courses on web design:
- a very lightweight one with very clear explanations and adorable
teacher, good for listening to while going for a walk :)
  https://www.coursera.org/learn/responsivedesign/
- a more comprehensive one about bootstrap 4:
  https://www.coursera.org/learn/bootstrap-4/
You might know these things already and even if not, you probably know
enough to come up with some basic styles for what we need; we don't
need perfect design anyway.

(b.3) Django (which you should know from inside out :)


PLANNING

(c.0) Open for discussion/suggestions ...
I would suggest that we create a new repository for your project under
MacPorts, you fork it and use your own fork for endlessly playing with
it, testing etc. (No need to fork though, you can just start
somewhere.) Once you think that part of your code is ready for review,
you submit a pull request, we review it and merge it. That way we
would end up with clean code upstream, all of it being properly
reviewed.

Once the code ends up in the MacPorts repository, that code should
basically be ready for deployment.

We use Trac as our main issue tracker. It depends on what others
think, but as a pilot project we could use GitHub issues & milestones
in this case. What do others think?

(c.1) I would suggest to turn your application (just plans, no
personal info) and spreadsheet with database into some
documentation/plan document in that repository. Then we can make
further suggestions for how to improve the plan before the summer
starts.

(c.2) Let's use the bonding period to open a bunch of tickets and
assign milestones to them, say:
  - "install MacPorts on X; milestone: community bonding"
  - "accept and store installation statistics; milestone: 1st midterm"
  - "deploy basic page; milestone: 1st midterm"
  - ...
  - "implement OAuth; milestone: stretch goals"
to match his timeline from the proposal. We might need to identify
further requirements (that could be implemented by someone else) for
the work to be complete (like: improve portindex2json, whatever needs
to be changed on the buildbot side, ...).

(c.3) Plan the API to get the data from the database in JSON format
(so that someone else could write an independent app with the same
display functionality).
More in a separate email, I guess.


MARKET RESEARCH

(d.1) Figure out if the app could be deployed somewhere temporarily. (Heroku?)

(d.2) Do some "market research" of different available graph drawing
toolkits and try to report pros/cons for our application. (Personally
I would like to find something that would allow zooming and sliding in
time, but even if that requirement is not met, that's OK.)


MACPORTS BASE

(e.1) Usually we would have suggested the students to get really
familiar with the Tcl and MacPorts base. I'm not sure how much Tcl is
really needed in this case. You might want to be able to modify
mpstats or portindex2json or something like that, so it helps to have
a basic understanding of our code, but the tricky stuff can be done by
someone else when needed.

Here's a video about Tcl from Clemens:
* https://youtu.be/46qshiDskrM

That said, you should become familiar with various concepts of how
MacPorts works from the user point of view, where all the resources
are (build infrastructure etc.), ...

PROJECT DESCRIPTION

The official project description still has some typos and could be
improved. It can still be modified until the coding starts. Low
priority, but let's think of some improvements during those three
weeks.


Mojca