Suggestions for my honours degree project...

[scotsman]

Would be interested to hear any suggestions people have for my honours project.

I had thought of attempting my own linux distribution, but I think this might be a little to much for me to attempt on my own.So if anyone has any other potential ideas for areas of study within *nix then they would be welcomed.

Thanks in advance.

[anomie]

Well, you posted under FreeBSD General, so expect to get general FreeBSD recommendations.

Honors project for what level and field of study? Is this a PhD in library science or a high school CS project?

[TerryP]

Personally I don't think anyone could offer a helpful opinion on that (let along with that much info).

You could always create an implementation 'libc' or a Linux LKM to support a piece of useful hardware, but that might not be your area of expertise.

[scotsman]

Thanks for the valid points.

It's a project as part of my honours degree in Computing.As for my area of expertise, most of my previous degree modules have been fairly diverse due to the limited selection on offer at my university.They have ranged from games development , project management in IT and Math of Space and Time.

Fortunately the unverisity has helped me further my education but the downside is that the lack of modules available has not exactly allowed me to specialise in any particularly area.

Off my own back I have been studying FreeBSD , administration mostly, and found it held my interest and I found it had a lot of scope for learning. So I though it may pose a good area of study for the project.

Maybe a better question would have been what area of FreeBSD would you recommend for a study project. Although in hindsight I think TerryP might be right in saying 'Personally I don't think anyone could offer a helpful opinion on that '.

Possibly I should revise the idea and focus on something else...hmm

[chavez243]

Custom Kernels: Useful Optimization or 1337 Perpetuated Mythology

Unix in the New Millenium: GNU, Torvalds and some geeks at Berkeley

(yes I know what GNU stands for...)

[ephemera]

Originally Posted by scotsman

I had thought of attempting my own linux distribution

Please don't.

how about a utility to build/upgrade ports similiar to portmaster but multi-process (or muti-threaded) so that ports can be build much faster on multi-core processors.

one option to speed up builds is to use make -j but that doesn't work with all the ports (or so i heard).

i thought of another approach:
let say we want to build port 'A'. 'A' will recursive depend on other ports. this gives a dependency tree rooted at 'A'.
Now scan all the leaf nodes and assign them to be build concurrently on the n cpus (leaf nodes don't have any dependencies). when a port build completes remove its node from the tree, then scan the tree for any new leaf nodes and assign it to an idle cpu. the process repeats until the root node 'A' is build.
thats the general idea, you could add other refinements.
or maybe you could come up with some other idea?

[TerryP]

make -j doesn't work with all ports, at least last time I tried it (FreeBSD ~6.0-release)

Originally Posted by ephemera

how about a utility to build/upgrade ports similiar to portmaster but multi-process (or muti-threaded) so that ports can be build much faster on multi-core processors.

An utility like portmaster that can make good use of SMP systems would be nice to have in the ports tree, especially if such areas of computer science interest you scotsman.

[scotsman]

Thanks for all the suggestions , i've still got a few weeks before I have to submit my project outline for approval so I will give it some serious consideration.

Cheers again