Version support for datastore/Proposal
- Name: Sascha Silbe
- Age: 29
- EMail: domain sugarlabs.org, user silbe (obfuscated for SPAM reasons)
- Wiki user name: sascha_silbe
- IRC nickname: silbe
- Native language: german (english preferred for technical communication)
- Location: Germany
- Work hours: about 12h to 00h local time, i.e. 10h to 22h UTC
- Open source projects (with other developers beside me) I participated in: Husky (Fidonet on Linux) (many years ago, co-founded the project; my profile on that page is rather outdated), OSM, contributed bug reports and patches to a large number of projects. Try a web search on my name, it's unique (AFAIK at least).
- Current affiliation with Sugar Labs: maintainer of the build infrastructure
About my project
- Name of project: Version support for data store / Journal
- Technologies used: The ones currently in use by the data store / journal. The bonus part might introduce additional ones (e.g. sqlite) for indexing.
Don't overwrite existing entries in the Journal with the same name (which currently means loosing the old content forever and happens automatically), but rather add a new version to the entry. Enhance the UI to allow easy (and simple to understand) access to "old" versions, including modification (which means automatically saving in a new branch). As "easy and simple to understand" isn't actually easy to implement, I'll concentrate on enhancing the current Journal view by adding previous/next buttons to the details view of each entry for the primary part of the project. Adding a version tree details view and possibly other ways of presenting versions are planned for the optional (based on remaining time) "bonus part". Metadata is going to be part of each version (and mutable without creating a new version) at first.
Version support for data store / Journal already was part of the original design concept for the Journal, but hasn't been finished yet.
There are several prototypes (Olpcfs, olpcfs2) that try to not only introduce version support, but also compatibility with legacy (i.e. non-Sugar) applications and, in the case of Olpcfs, support for the activity-based view. While those designs (and partial implementations) are quite interesting from a research point of view and might prove useful later on, the issue of legacy application compatibility is a rather large can of worms, esp. if the applications are supposed to have access to multiple versions (there are other issues as well, e.g. treatment of multiple files written by the same application, deletion, handling of file names and file types).
For this reason, I don't believe that any of these prototypes will mature enough to get integrated in Sugar even mid-term. My project focusses on the version support instead, enhancing the current data store instead of replacing it, with the Journal as the only intended user of the versioning API (though, at least in theory, regular activities could access it as well).
The details of the project (i.e. on-disk format, exact API and UI design) needs to be considered with the Development Team resp. Design Team, but the general approach is quite clear.
- Application deadline
- Easter (sunday); UI mockup submitted for review by Design Team
- start of (university) term; announcement of accepted GSoC proposals
- submitted API draft for review by Development Team
- SugarCamp Europe 2009
- start of GSoC
- current code examined and understood; API, on-disk format and UI design chosen
- data store enhanced to be able to deal with versions (basic API)
- added (working) prev/next buttons to Journal details view
- added support for importing from existing data store
- added unit tests (and potentially regression tests), fixed all known bugs, submitted for review by Design Team
- GSoC midterm evaluation ("working and 90% done"); added indexing (e.g. using sqlite)
- code integrated upstream for increased exposure (testing!); started discussion on extended UI design (version tree etc.)
- end of (university) term
- end of GSoC
- Fedora 12 release; Sugar 0.86 release short time later?
Me and the community
Action after getting stuck
For this project, getting stuck means needing advice on UI issues. As I can ask both the whole Design Team and a friend of mine for input, it's rather unlikely that nobody will be around for a significant amout of time. Also I don't see the code produced by this project as the one and only answer to the problem, but rather as a start of a process. So I could just do an arbitrary choice and continue with the project, it's still adaptable later.
Impact of this project
One of the initial design goals will actually be provided. Instead of giving users the choice of irrevocably deleting either the old or the new content upon exit as current desktop applications do (unless the user explicitly invokes a "Save As" operation prior to exit), the computer will now retain both and provide the user with ways to access them.
While I would very much like a Pilot to occur in my vicinity (once both this project and the Rainbow support have been merged), I'm not yet sure what the best target would be as I'm not directly involved with any school.
Keeping the community updated
For regular updates, I'll probably update a wiki page, as those interested in it can subscribe and get notifications via email. There are going to be several projects going on, so sending the reports to the regular mailing lists feels like spamming. For problems and questions, I can use both the IRC channel and the mailing lists, depending on the exact nature and target audience (e.g. details on the current data store are best asked on IRC, while UI design questions are best elaborated in a mail thread where there's sufficient time to do research to back up arguments).
I'm skipping the screenshot of the simple text replacement task for time reasons. I guess being the maintainer of the build infrastructure and having committed several patches on the bugtracker should be sufficient substitute. :)
- T-Shirt size: depending on the actual size (instead of just what's printed on the label) it's M up to XXL. Usually I go for L.
Great learning experience as a child
Wow, good question. I don't remember any particular event right now. In general, I learned most (and most easily) when I was actually doing something, without any external help. Reading books (even text books, though not the school ones) was fun as well and helped me get further insight and new ideas. That's still the case, though "the internet" has replaced books for me.
Anything else to like the project more
I don't think there's any need for raising the priority of this projects even higher, but as you asked I'll give a reason nonetheless: Working on this project will significantly increase my understanding of the data store, the Journal and the whole Glucose software stack. As a result, adding Rainbow support is going to get a lot easier.