[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]


Matt Porter <mmporter@home.com> writes:

> On Mon, Nov 08, 1999 at 05:34:50PM -0500, Michael Stone wrote:

>> When I asked about freezing three months ago, the main criticism
>> was that we didn't have working boot floppies. Now you're saying
>> that the reason we can't use the old boot floppies is that they
>> don't support powerpc? If that's the only hold up, we could have
>> released potato already and frozen *woody* in 2000. (IOW, if a new
>> arch isn't 100% ready it seems a whole lot more reasonable to not
>> put it in until it's done, rather than holding up n other archs.)

There are tons of problems with the slink boot-floppies, including 5
RC bugs which should be closed in the 2.2 series, not to mention all
the changes coping with the new potato base system, which itself is
still riddled with bugs.  One thing people think is that most
boot-floppies work goes on in the boot-floppies CVS area.  That is not
quite correct.  A ton of work has to go into working with the
maintainers of other base packages, getting them to fix bugs (such as
just fixed in pcmcia-cs).  The fact is that maintainers who have
packages in base don't common work in a base only system.  No-one
does.  So it's up to the boot-floppies team (and testers) to sort all
this out.

Moreover, working with all the architectures to make sure that stuff
is ready, and with the kernel-image maintainers, for all
architectures, is extremely time consuming.  In fact, you might not
realize it, but we got a late-model 2.2.x sparc image only last week.
Some architectures still don't have approriate kernel-images.

We've close about 80 bugs in the slink boot-floppies, too.

> When I started working on boot-floppies, I worked on it for at least 3-4
> months doing powerpc porting work before I ever once saw an x86-based
> developer show up (excluding Adam who was doing plenty of documentation).

Yes -- Enrique was our i386 guy.  He basically has been pretty absent
over last spring and summer.  He officially polled for a new
boot-floppies manager in July.  I think out of desparation rather than
particular talent, I picked up this duty in last August-ish.

> I've fixed a ton of problems generically across the various archs as well
> as doing a bunch of powerpc specific fixes.  What I'm telling you is
> that potato is supposed to have the powerpc arch as well as some value add
> in the boot-floppies that include dhcp support and ftp/http installation.
> The x86 boot-floppies do not work either in the current CVS, and so this
> delay is the fault of lack of developers for x86 boot-floppies. 

Geeze, Matt, let's not point the finger.  You're just getting everyone
into hot water.  The truth is that issues across the board have caused
the delay.  In fact, as I said before, we've spent more time waiting
for packages to be ready than in trying to get our hacking done
(pointerize, kernel-images, perl-base).

> Had we decided to not allow in an updated busybox plus the new
> installation features, we would be ready for the freeze.

No way.  We still need a new profile/task selection mechanism.  We
wanna get debconf into base for the apt configurator functionality.
No one is really working on bootp/dhcp.  I don't consider us "freeze
ready" until we're feature complete and reasonably tested.  It's true
that busybox has caused us at least a 2 week delay, and prevented the
release of 2.2.2 thus far.

> I hate to start an arch-war but trying to point fingers at a non-x86
> arch as holding things up is about as far from the truth as you can
> get.

I agree with you there, but better to not point the finger at all.

.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>

Reply to: