Re: where are graph algorithms used under an operating system project?

[Dan LaBell <dan4l-nospam%verizon.net@localhost>]

Testing the final product.

Especially, if you're considering, it scientifically, then operating  
system is only half the architecture.

Since, you say project. then you can mean something that includes,  
what would be an application on
on something like netbsd.  Because netbsd and other unix like -- I'm  
going to interrupt myself, to
say that Painting algorithms are part of Graphs, as far as the  
science goes -- Anyway, presently, the only
way to absolutely test for pre-fail, in a fixed disk, that I can use,  
is a program, that I'm constantly
tweaking, that began, as my stab, at progress, since at that time it  
was hard to grab, if you can believe
that.   Anyway, something that painted the directory tree, while  
looking for slow spots, would be better,
than an offline, or post backup:
#./mbps < /dev/rwd0e > /dev/zero 2> log & sleep 2.5 : tail -f log
The trick is to have it print out something like TOO SLOW!, and the  
end of line, when it's slower,
than a limit, like 10 MB/s.  If it's less 1, it's failing now.
BTW to zero is necessary, null would just testing the seek function.

So, because netbsd has the raw device, I can do this, without having  
use the Waterfall Model.

On Jun 18, 2019, at 3:28 AM, Frédéric Fauberteau wrote:

> Le 2019-05-17 09:41, Mayuresh Kathe a écrit :
> > due to a lack of background in formal computer science, i was  
> > under the
> > impression that graph algorithms are only useful in networking  
> > related
> > applications/areas. just noticed that even scheduling requires  
> > some level
> > of graph algorithm knowledge. am i right in my reading/ 
> > observation? if not
> > which areas of an operating system require a good grasp of graph  
> > algorithms?
> > thank you.
>
> FMHO, an operating system has to be efficient and maintainable. And  
> I am not sure it is a reasonable idea to implement graph algorithms  
> (in a general way) in OS. But special cases of graphs as lists or  
> trees are commonly used (e.g red-black trees had been used to  
> implement a scheduler on Linux). When you talk about graph  
> algorithms, it sounds to my ears as Dijkstra or Ford-Fulkerson, and  
> I am not sure these algorithms are implemented in OS. But sure that  
> lists and trees are used for schedulers, filesystems and probably a lo