Re: Installation of FreeBSD 8.0-Current-2009-amd64-dvd

On 2009-May-12 01:47:49 -0400, Mehmet Erol Sanliturk <m.e.sanliturk_at_gmail.com> wrote:
>Personally I would be very happy to be able to contribute to development of
>FreeBSD as much as possible . Up to now I could not be able to establish a
>working system as a network which is suitable for my needs , but I am
>working toward this .

Your efforts are appreciated, but as others have indicated, you have
chosen an approach that requires a steeper than necessary learning
curve for yourself.

>As an example , I want to mention other operating systems : Debian : When an
>update is available an icon is appearing and saying that an update is
>available . When it is accepted the update is applied .  At present my
>Debian 5.0.0 is up to date in that way .

I don't believe that model is appropriate for FreeBSD.  Firstly, by
default, FreeBSD does not have a GUI (installing X11 is optional) and
many FreeBSD installations are servers that don't have either screens
or logged in administrators.

Secondly, what do you define as an "update"?  There are maybe 100 port
updates each day and about the same number of updates to -current.
Do you really want to get several hundred popups each day?  If so, you
might as well just subscribe to the relevant commit mailing lists.

>When a part is modified and it requires a test , all of the participating
>related systems may be notified by the automatic testing system .

This is a nice idea but I don't think it's practical.  As has been
mentioned, the -current snapshots are just whatever is in the tree
when the build triggers.  It may contain an arbitrary number of
changes from the previous snapshot and be in the middle of a set of
related commits.  How do you determine the change list to present to
the user?  How do you handle it if the user has skipped a previous
update (which means the list of changes from their running system
is the union of several snapshot sets)?

>The only requirements will be that tests will not crash the complete
>installations at to the points that the system will not be even able to boot
>any more which it will not be able to send any message back about test
>results .

Whilst -current is generally OK to use, it is quite possible for it to
eat your filesystem if you update at the wrong point.  The only safe
way to run -current is either to use a throw-away virtual host or to
follow the mailing list and understand what has changed and what is
being worked on.

>As an example : In one of my Linux computer , to reclaim file space
...
>making the system completely unbootable , unable to rescue current contents

That sounds like there was a serious bug in the dependency information
and seems more a reason not to implement your proposal.

>hard disks are large per drive or a dedicated hard drive around 160 GB is
>around US $ 50 ( US dollars ) which I think many users would use such an
>option .

And for many other users wouldn't be an option due either to cost or
physical space (eg, I can't fit a second HDD into my laptop).

>In a multi-boot system , booting the 8.0 Current and applying tests and send
>reports back may not be difficult for the participating users ( at least
>they are willing to apply such steps ) .

I find rebooting annoying - I keep a lot of state in xterms, browsers
and editors.  Rebooting means losing all that so I avoid rebooting
unless I really have to.  A session manager can restore my 50 xterms
and 4 or 5 emacs but it can't restore all the less(1) sessions with
various marks set.

>The hardest problem is to have  such a testing environment . The present
>model is difficult to
>participate . It is very flexible , but requires much knowledge . To make it
>a more structured
>will reduce required amount of knowledge but will increase possible
>participation and back contribution .

What impact will reducing the flexibility of FreeBSD development have
on the Project?  Most FreeBSD developers are volunteers and if you make
things too hard for them, they'll just leave.

If you believe you can come up with a workable solution, I suggest you
come up with a more detailed proposal and discuss it on -arch.

>Writing specifications and requirements and then opening it to academic
>environments may attract interest .

If you know people who want to write specifications, feel free to point
them at the FreeBSD project.  For things in the various ideas pages,
there are contacts who should be involved.

-- 
Peter Jeremy