| 1 |
This is final report for Project Grumpy regarding Google Summer of Code |
| 2 |
|
| 3 |
|
| 4 |
Now that the official 'Pencils Down' date has been reached, it is time to look |
| 5 |
back and sum up the things that were done and what was not. |
| 6 |
|
| 7 |
Initial plan |
| 8 |
============ |
| 9 |
|
| 10 |
The initial plan listed following deliverables: |
| 11 |
* Database storage for ebuild metadata (done) |
| 12 |
* Tools for gathering and managing metadata (done) |
| 13 |
* Upstream information checks (mostly done) |
| 14 |
* User-interface tools (mostly done) |
| 15 |
|
| 16 |
Project evolution |
| 17 |
================= |
| 18 |
|
| 19 |
Although I had a clear plan in the beginning of the project, there were |
| 20 |
several hurdles that forced me to make several big changes to the overall |
| 21 |
design of the Grumpy project and for few times brought the whole project to a |
| 22 |
complete halt. |
| 23 |
|
| 24 |
As my first approach was to use MongoDB's document-oriented storage engine, |
| 25 |
devs handling Gentoo's infrastructure tasks were not too happy to support |
| 26 |
one more database on their machines and also my mentor and other developers |
| 27 |
who were a bit more involved with the project pointed me to several failures |
| 28 |
with Mongo. So eventually I went with PostgreSQL using SQLAlchemy as |
| 29 |
high-level ORM acting as middle layer between database and application. |
| 30 |
|
| 31 |
Now getting the data model and database working was tricky and I still believe |
| 32 |
that there will be a lot of upcoming changes regarding the database schema. |
| 33 |
|
| 34 |
Unimplemented features |
| 35 |
====================== |
| 36 |
|
| 37 |
The first unimplemented 'big' feature that I could not implement was the |
| 38 |
'30-day stabilization checker'. This is hard to implement. Period. |
| 39 |
|
| 40 |
Secondly, the JSON-API and commandline utils for developers. This was because |
| 41 |
I got more and more stuck with my deadline and my Grumpy backend really did |
| 42 |
not have required features for commandline tools until it was too late. But |
| 43 |
this will be one of the first features that will be implemented post-GSoC. |
| 44 |
|
| 45 |
I did not implement LDAP-authentication backend and simply went with JSON and |
| 46 |
@gentoo.org email check. I just went with simple enough and working solution. |
| 47 |
|
| 48 |
I also wanted to implement Tinderbox reports support, but due to two bugs I |
| 49 |
ran into on my desktop machine (e1000e NIC hang (kernel) and carrier detection |
| 50 |
issue (dhcpcd)), I had to drop this idea after wasting considerable amount of |
| 51 |
time. |
| 52 |
|
| 53 |
Implemented features |
| 54 |
==================== |
| 55 |
|
| 56 |
Grumpy backend that handles syncing the portage contents with database and |
| 57 |
web interface on top of it. Three QA-plugins: developer and herd checker and |
| 58 |
upstream version bump check plugins for packages from GNOME and PyPi. |
| 59 |
|
| 60 |
Future plans |
| 61 |
============ |
| 62 |
|
| 63 |
I really want to get Grumpy up and running using Gentoo's infra. So far I have |
| 64 |
not had much luck dealing with infra. And of course get bug #332623 solved. |
| 65 |
|
| 66 |
And near future todo list regarding Grumpy itself: |
| 67 |
* Implement tinderbox plugin which is a huge step towards better QA. |
| 68 |
* CLI tools using JSON-API |
| 69 |
* Improve Web UI usability |
| 70 |
|
| 71 |
|
| 72 |
|
| 73 |
Päikest, |
| 74 |
Priit :) |
Archives