[openbeos] Re: too much focus on kernel? baloney!
- From: "Daniel Reinhold" <danielr@xxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Sat, 03 Aug 2002 16:07:54 CDT
>If I may be so bold: USB is something I'm so keen to work properly
that I'm
>prepared to offer some money as an incentive to get it done - contact
me
>directly if you're interested. There is definitely too much focus on
the
>kernel at the moment. Is this because it offers the most challenges or
>glory for programmers?
>
>Charlie
No, there is not too much focus on the kernel. Not even close.
Ok, several people have made statements like this lately and it is
starting to get my goat, so I've got some of my own comments to make...
I wrote up a news item awhile back (kernel kraziness) that made it
appear as tho *everyone* was working on the kernel. My only intent at
the time was to make it clear to the general public that the project
hadn't died. The website wasn't being updated much (largely due to new
webmaster Kurtis working on the backend) and the team mailing lists
weren't having any activity. But there was a big flurry of activity on
the kernel and I wanted to get the word out on that.
Unfortunately, a number of people have taken this too literally and
have concluded that the kernel is receiving too much attention while
other components are being ignored. This is totally false.
The truth is, the kernel is FINALLY receiving the attention is has
badly needed for a long time. Until the fork, nothing (essentially) was
being done on the kernel -- and that was not a good thing. Michael did
not want to fork too early while Travis was still actively improving
NewOS. Trouble is, Travis (who got very busy with other things) stopped
working on NewOS for awhile. And even when he resumed work on it, his
new stuff mostly moved NewOS into areas that weren't usable by OBOS.
IMO, the fork should have occurred months earlier.
In any event, the fork finally happened and this left *A TON* of stuff
that needed to be fixed immediately. There is a quite non-trivial gap
between the APIs used in NewOS and those needed for OBOS. In some
cases, it's merely a matter of changing function names; in other cases
(e.g. VM) it involves a complete re-write.
So the kernel very badly needs works and lots of it. We will never get
to R1 w/o a fully working and robust kernel and we're still quite a
ways from that. We could have 10 more people join in on working on the
kernel and that would not be overkill at all.
Another thing to remember is that those working on the kernel were
either already on the kernel team or had immediate needs from the
kernel for their kit. For example, Axel basically has BFS all but
finished. He is not abandoning that kit by working on the kernel. In
fact, BFS can't even be used by the NewOS kernel. So he began the
necessary work of converting the VFS layer to make it compatible with
the BeOS API and make it work with BFS. Likewise, David Reid had a
number of requirements from the kernel in order to integrate the net
stack. He had to spend weeks just changing function names, header
files, etc. just to get the darn thing building and approaching
something that resembled what the net stack could use.
We are also lucky that a bright and energetic chap named Angelo Mottola
came our way and started working on the kernel. His addition did not
remove work from some other kit. As it is, very few of those were
actually listed as 'Kernel Kit' members have even been heard from since
the fork.
In the mean time, others continue their own work. Erik is still working
on the Interface Kit, BMessage, etc. DarkWyrm is still working on the
app_server. Marcus is still working on the Media Kit. Mike Wilbur is
still working on the Translation Kit. Ingo and Tyler are still working
the the Roster stuff, which is the blocking point for finishing up the
Storage Kit. Phil is still working on the DUN preferences. And others
are still working on their respective apps.
It is probably true that more work (in terms of checkins to CVS) is
being done on the kernel than on the other kits at this point in time.
It is GREAT NEWS that this work on the kernel is being done because it
is so needed and so vital. The bad news is that more work isn't being
done on the other fronts.
It just so happens (largely coincidence really) that the majority of *
actually active* developers on the project happen to be some of the low
-level coding types. Do we tell Axel, Phillipe, Francois, David, etc.
to go work on other userland stuff because it looks like too much
inbalance on the kernel side? Ridiculous! We need *MORE* kernel
developers, not less. We also happen to need a lot of programmers for
the other kits as well.
It is a wonderful thing that the kernel is now receiving proper
attention. I didn't hear anyone complain for all those months that the
kernel was not being worked on... that is was receiving "too little"
attention, even tho that was exactly the case. So there is a helluva
lot of catching up to do. You should be very, very happy that this
catching up is actually being done!
Other related posts:
