Re: Packages for Current ( 10.0 )

On Sat, Feb 18, 2012 at 12:11 PM, Mark Linimon <linimon_at_lonesome.com> wrote:

> On Fri, Feb 17, 2012 at 12:04:45AM -0500, Mehmet Erol Sanliturk wrote:
> > I would be very happy I can contribute anything to development of
> FreeBSD .
>
> We are always happy to have help :-)
>
> > If there appears an agenda of testing problems and explicit instructions
> > how to apply tests , me and other persons may apply them and report the
> > results .
>
> Well, there really isn't.  There are some regression tests for src, but
> we have never established a framework to run them all automatically.
> Perhaps this is something you might be interested in?
>

I am going to prepare a message like a specification for testing framework
and send to you .
In that framework , I want to adhere the current FreeBSD development ( web
sites , available ports / packages , usability of existing testing
facilities ) .

I am planning to write the specification in such a structure that when a
person takes it he/she will be able to generate an applicable project from
it . This is not difficult for me because I was a computer science
instructor in the University .


>
> > In a message ( I do not remember its author's name  ) it is said that
> there
> > is no a farm of FreeBSD testing machines .
>
> I presume that's "now" for "no"?
>
> There are several sets.  Here's how they are set up.
>
>  - there is a "src tinderbox" which continually rebuilds the FreeBSD
>   src tree, for various combinations of architectures and osreleases.
>   (For src, the architectures can be cross-built.)  These are intended
>   to sanity-test that src is still buildable; in general the resulting
>   binaries are not made available.
>
>  - there is a "clang buildbot" whose purpose is to build FreeBSD src
>   under clang continuously.
>
>  - various people maintain "ports tinderboxes".  These are optimized
>   for test-builds of one or at least a subset of the ports tree.  In
>   general the resulting binaries are not made available.
>
>  - there is a new effort, Redports, to assemble a collection of ports
>   tinderbox machines and make them available to interested people.
>   We are actively working on this.
>
>  - portmgr maintains the "pointyhat cluster" that do the package builds
>   which are uploaded.  These are optimized for building the entire
>   ports tree in a secure fashion; the resulting binaries are made
>   available.  We are in the process of getting more machines online.
>
>  - the pointyhat cluster is also used for "-exp runs" where portmgr
>   regression-tests proposed changes to the overall ports tree to
>   try to ensure as few regressions for large changes as possible.
>
> > If we can generate such a testing ecological system , I think , FreeBSD
> > development will benefit from it very much .
>
> I agree.  But, for src, it's not something that I know much about,
> and will have to defer to others to comment.
>
> mcl
>

At present , there a very valuable efforts for testing FreeBSD as you
explained above .
My approach will be not only testing the correctness of compilation
but also execution correctness .

As an example ,

when a snapshot is downloaded , installed and tried to boot , even it is
NOT booting .

My goal is to prevent such and other execution failures because every
failure is a waste of very valuable human time and other resources .


My primary profession ( university graduate subject ) is
Mathematics/Statistics/Operations Research .
During my undergraduate study I took also many electives from Industrial
Engineering such as "Motion and Time Study" which its subject is to design
"work procedures that consumption ( such as time , energy , etc. ) is
minimum while the outcome ( the amount of work performed , completed ) is
maximum .

>From these points of view , I think , there are possibilities to improve
development and wide adoption of FreeBSD which is a direct contribution to
humanity welfare .


Thank you very much .


Mehmet Erol Sanliturk