Re: Tool to find dependencies precisely

[Marc Espie <espie%nerim.net@localhost>]

On Tue, Jul 14, 2015 at 04:42:53PM -0000, wca%mail.bg@localhost wrote:
> 
>    Dear Sir,
> 
>    I have a profound interest of the UNIX systems and particularly the BSD
>    flavour. I found it is the most beautiful of all UNICES. It is nice
>    occassion to learn from NetBSD website, you're looking for developers
>    for various projects. Some of the projects are very challenging. I can
>    solve probably all of the hard ones in a unique manner.
> 
>    I would like to work on a tool for building a dependency graph
>    precisely. Building a good quality tool for such a purpose is not a
>    trivial task. Why the task is not eligible for funding?
> 
>    I implemented some graph algorithms in the past therefore I know the
>    matter in detail: how to create optimal structures and algorithms.
>    Programs I have developed were related to solving large sparse linear
>    systems. Problem of graph partitioning were crucial in these methods.
> 
>    Precisely to formulate, my inquiry is to become a permanent developer
>    member on the more serious parts of the system but for some
>    compensation (for example payment per hour). You will have a benefit of
>    a good developer, who at some point may work even on whole system. My
>    work could add a significant value to the BSD system. I would describe
>    myself not simply as a good coder but also as a person with original
>    conceptual attitute.
> 
>    If you're interested in my application I may send a short resume of my
>    previous work. Program for graph path also needs some details (what
>    input, output to be used etc., how to be a more general tool).
> 
>    Thank you for your attention! I am looking forward your letter!
> 
>    Best regards,
> 
>    vsa

What part do you think is missing in the BSD ecosystem ?

- complex dependency solver -> pkgng has got that. Talking with bapt%freebsd.org@localhost
on various occasions, I know he explicitly funded someone to have the same
level of dependency handling that various linux distribution tools have.

- dependency handling to build ports -> every single BSD has reasonably
good tools to do that.  It doesn't require very sophisticated algorithms,
actually. And the size of the graph isn't large (we're talking on the
order of 10000 nodes in a rather sparse graph).

- discovery of missing dependencies. That part can probably be improved, but
I'm pretty sure that's not a graph problem at all.  Scanning thru configure
script output and looking at dynamic libraries linkage doesn't look like
a graph problem.