Ask Slashdot: What Does the FOSS Community Currently Need?

First time accepted submitter d33tah writes "In the summer term of my final year of IT's bachelor's course in my university, every student is obliged to develop his own project; the only requirement is that the application would use any kind of a database. While others are thinking of another useless system for an imaginary company that nobody would actually use, I'd rather hack up something the FL/OSS community actually needs. The problem is — how to figure out what it could be?"

Statistics

[jevring]

How about something fun, like filesystem statistics? Keep track of the most used files to make sure you spread the disk and your mental load equally. Quite possibly useless, but could be fun to do. The hooks into the FS might be the hardest part about this, though.

Write a generic ETL app. Quite useful. Might be many out there, though. Probably few good free ones..

Or something that converts a (well known) log format into database entries for the purpose of easier statistics than what grep can provide?.
For instance, take a webserver log, dump it into the database and generate something like a visitation path..
The database isn't technically needed for this, of course, but with a large dataset, you can't keep it all in memory, so it would be useful..

Re:What do YOU need.

[azalin]

What about an MP3 Database over removable drives, that lists duplicates and relatives (ie copies or same song - different sample rate, or same song live - studio version), allows mass renaming/updating (or even auto labeling from an online source) - Bonus feature: remember update for other copies of the file on external drives. Along with the option to label drives as "backup of x" (needs all songs in x), player (files may be removed if they exist elsewhere), storage or import (no changes to drive). Add a timestamp for last sync in the database for each drive.
This might not be the holy grail, and may even exist(?), but it would be useful and is the first thing that came to mind.

OpenStreetMap

[Max_W]

Mapping tools for mobile devices. Like "OSMPad", but better, if it possible at all.

bioinformatics

[tloh]

If your comfort zone can be stretched into biotechnology, there are many opportunities for analyzing huge volumes of data in genomics/proteomics. As one modest example: a select number of model organisms are commonly used for basic research. Is it feasible to build an app/tool that can gauge the suitability of an experiment subject for a particular scientific inquiry based on available genomic data? Recently, I heard a talk by a researcher in autism attempt to find a mouse model of the disorder based on observed behavior in cognitive experiments across many different laboratory strains that have been inbreed to very exacting parameters for other experiments. Given the level of detailed information on these particular strains, it is easy to see how convenient it would be to have a tool that can mine their genomes for a particular trait or set of traits or perhaps even do an in silico genetic engineering experiment before any resources are physically committed. Even if hardcore biology isn't your forte, you might maybe talk to someone who teaches the subject and ask what tools can be developed to help visualize or otherwise communicate conceptual information that derive from databases of the type kept by organizations like NCBI [nih.gov].

copyright

[shentino]

Double check your university's policy on copyright of student work.

Peer Code Review Software

[Cory Fisher]

Write a cross platform easy to set up and use peer code review software (take hints from CodeCollaborator theirs is good but expensive). It should have hooks into Git/SVN and be easily extended in the future to include other version control systems. It would also be cool if it had the ability to have source code scanning plugins like phpcs (code sniffer) or phpmd(mess detector)... I'm a PHP guy you can tell, but I'm sure the guys from the java, c++, and other communities could use similar tools. Make sure it has an easy to set up web interface (you could package a webserver into the deal that listens on whatever port is configured during the setup process).

Re:You should think of what your teachers expect

[LSDelirious]

Instead of guessing or tape measuring remaining filament lengths, is there some reason you can't weigh the remaining materials? Seems like the filament would be a consistent diameter and density that with a scale of decent precision you could weigh a known length and have a fairly accurate idea of the remaining length.

Re:Hmm...

[theVarangian]

Better hygiene. Less beards. More women.

I'll wash off the stink and you can swamp me with women but you'll have to shave my manly beard off my pale dead face. There is no way you'll get us beardy weirdys to participate you strange metrosexual war on body hair, it's donwright unmanly.

Matching contributors to needs

[Anonymous Brave Guy]

If the OP really means what the community as a whole needs rather than one useful thing for part of that community, then ironically I think you've just nailed it: more than anything else, the community needs a way to match up willing and able contributors with projects that could benefit from their contributions.

To do that, the OP could develop a simple database that understands things like:

• different kinds of contribution ("I want to help with programming")
• technical skills ("I program C++ pretty well, and a bit of Ruby")
• application domains ("I like graphics-related projects")
• levels of difficulty ("this is a million-lines project" => it will take a while to get into and might need significant infrastructure installed to work on it)
• availability ("I can spare an hour or two a week" => probably better to help with small things on smaller, more accessible projects).

Provide some sort of keyword store (extension: recognise related entries/common aliases) or defined scale for each property, let projects say what they need and volunteers say what they're willing to contribute, and help people get matched up.

This has the handy advantage for the OP of being readily scalable from a simple proof of concept with a simple native or web-based UI right up to a full-blown and genuinely useful service if you can find a way of getting it hosted properly. It might help particularly with contribution in areas other than programming, which in practice is often where OSS projects run by volunteers for free start to fall behind commercial projects run by businesses with cross-disciplinary teams.

Re:Matching contributors to needs

[jevring]

It's strange that such a site doesn't already exist. There are a bunch of hire-a-freelancer sites out there, but nothing for something like this. If you build it, I will come! :D

How about a replacement for Slashdot?

[water-and-sewer]

Semi-serious. I think Slashdot's got one of the best content/comment/moderation systems around - certainly better than Reddit, way better than the ashes of Digg, and more useful than Usenet.

Build a FOSS database with whatever improvements you design, as the underpinnings for a new Slashdot not owned by some mega-corporation intent on shoveling crap articles at us, like "how to get employed by RedHat" or video interviews about random horse crap?

A database filesystem

[pr0nbot]

Write a (Linux, BSD) filesystem driver that keeps its file metadata in a database.

Use queries to construct the filesystem layout. E.g.

• /bin -> files where executable=true and package=LSB (or whatever)
• /sbin -> files where executable=true and package_owner=root (or whatever)
• /usr/local/{name}/ -> files where package={name}
• /etc/{name}/ -> files where package={name} and type=configuration

...and so on. Don't ask me what the exact queries should be - the idea is just that files are arranged in the filesystem because of their attributes rather than having a single home.

Add a chattr command (or somesuch) to modify metadata for a particular file, or implement the inverse of the queries as attribute changes (i.e. mv /bin/ls /sbin/ls causes the owner=root attribute to be set on the file).

I'm not saying it'd be useful to anyone in the FOSS world, but it would be great fun.

OS/2 Clone

[martiniturbide]

The OS/2 community is in need of open source clone for our beloved OS.
We have partially open source components on all the layers, but some need to be finished and glue them together.

We need:
- Workplace Shell replacement (xWorkplace can be used)
- SOM replacement (FreeSOM can be used)
- OpenDOC (docshell)
- PM (Presentation Manager) replacement (FreePM can be used but is missing a more)
- OS/2 Kernel replacement
- TCPIP replacement.
- Drivers
- OSFree project code can also be used.

RTFS

[Anonymous Coward]

How many people here can't read the fucking summary? here's a shot at it. How about a de-duplicator for music/photos that would (nicely) hunt for media, throw the metadata in the database, search for identical and almost-identical files, and then beautifully show the output. Bonus points if you beat the standard interface to these things which is just a list of duplicated files. I'd suggestb bubble diagrams that show how many files in which folders are duplicates of others.