Re: MIPS port backlog, autobuilder machines and some arrogance

[Matt Zimmerman]

On Thu, Nov 20, 2003 at 12:54:07AM +0100, Goswin von Brederlow wrote:

> Matt Zimmerman <[EMAIL PROTECTED]> writes:
> > The whole point of signing packages is that it is not anonymous at all, but
> > traceable back to the signer.  Assuming the keyholder protects his key
> > adequately, there is reasonable assurance that the keyholder and the signer
> > are the same person.
> 
> Exactly my point.
> 
> As a non DD running a buildd I have much more and anonymous access to
> packages being build. I and some others are aparently trustworthy
> enough by their DD friends but not by the DAM.

The burden lies with whomever is doing the signing.  They are accepting
responsibility for what they upload, and if that involves trusting you, then
they are taking responsibility for you as well.

-- 
 - mdz

Re: MIPS port backlog, autobuilder machines and some arrogance

[Goswin von Brederlow]

Matt Zimmerman <[EMAIL PROTECTED]> writes:

> On Mon, Nov 17, 2003 at 03:56:44PM +0100, Goswin von Brederlow wrote:
> 
> > DDs have to sign and upload a package with a backdoor.
> > 
> > On the buildd I can install a gcc or other tool that will silently add
> > a backdoor to anything getting compiled and the buildd admin will sign
> > and upload the package for me.
> > 
> > Much more anonymous.
> 
> The whole point of signing packages is that it is not anonymous at all, but
> traceable back to the signer.  Assuming the keyholder protects his key
> adequately, there is reasonable assurance that the keyholder and the signer
> are the same person.

Exactly my point.

As a non DD running a buildd I have much more and anonymous access to
packages being build. I and some others are aparently trustworthy
enough by their DD friends but not by the DAM.

MfG
Goswin

Re: MIPS port backlog, autobuilder machines and some arrogance

[Joerg Wendland]

Martin Michlmayr - Debian Project Leader, on 2003-11-19, 14:32, you wrote:
> * Ingo Juergensmann <[EMAIL PROTECTED]> [2003-11-16 15:40]:
> > > Yes, a fairly powerful machine has recently been donated to Debian and
> > > we're currently working out where to host it.
> > 
> > Where is it located?
> 
> In the States; not really worth shipping to Germany.  However, I'll
> see whether we can find some nice systems in Germany as well.

As I already said at LinuxTag in Karlsruhe earlier this year, my
employer is willing to host machines for Debian in Germany.  So if need be, 
I can offer rackspace and connectivity in Ulm or Frankfurth.

Joerg

-- 
Joerg "joergland" Wendland
GPG: 51CF8417 FP: 79C0 7671 AFC7 315E 657A  F318 57A3 7FBD 51CF 8417


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Ingo Juergensmann]

On Wed, 19 Nov 2003 19:20:09 +0100, Guido Guenther wrote:

> On Wed, Nov 19, 2003 at 06:43:00AM +0100, Ingo Juergensmann wrote:
>> Perl is building fine as well now on mips, although it is marked as
>> not-for-us. See
>> http://m68k.bluespice.org/cgi/package_status?mips_pkg=perl&searchtype=go
> This might be due to the fact that the autobuilders don't run recent
> enough kernels. I offered to binary NMU perl at least two weeks ago, but I
> was told that this will be taken care of soon. Cheers,

Well, when it would be a real kernel issue, I wonder why I was able to
successfully built perl 5.8.2-2 on 2.4.19-r4k-ip22? ;-)
But I think we have either to wait until the buildd maintainer takes care
of it or lolando finishes building perl on casals.d.o. ;-))

Ciao...
  Ingo

Re: MIPS port backlog, autobuilder machines and some arrogance

[Guido Guenther]

On Wed, Nov 19, 2003 at 06:43:00AM +0100, Ingo Juergensmann wrote:
> Perl is building fine as well now on mips, although it is marked as
> not-for-us. See
> http://m68k.bluespice.org/cgi/package_status?mips_pkg=perl&searchtype=go
This might be due to the fact that the autobuilders don't run recent
enough kernels. I offered to binary NMU perl at least two weeks ago, but
I was told that this will be taken care of soon.
Cheers,
 -- Guido


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Matt Zimmerman]

On Mon, Nov 17, 2003 at 03:56:44PM +0100, Goswin von Brederlow wrote:

> DDs have to sign and upload a package with a backdoor.
> 
> On the buildd I can install a gcc or other tool that will silently add
> a backdoor to anything getting compiled and the buildd admin will sign
> and upload the package for me.
> 
> Much more anonymous.

The whole point of signing packages is that it is not anonymous at all, but
traceable back to the signer.  Assuming the keyholder protects his key
adequately, there is reasonable assurance that the keyholder and the signer
are the same person.

-- 
 - mdz

Re: MIPS port backlog, autobuilder machines and some arrogance

[Andreas Metzler]

Ingo Juergensmann <[EMAIL PROTECTED]> wrote:
[...]
> Perl is building fine as well now on mips, although it is marked as
> not-for-us. See
> http://m68k.bluespice.org/cgi/package_status?mips_pkg=perl&searchtype=go

> paco:/home/ij# ls -l *.deb
> -rw-r--r--1 root root35172 Nov 18 22:24 
> libcgi-fast-perl_5.8.2-2_all.deb
> -rw-r--r--1 root root   651210 Nov 18 22:32 
> libperl-dev_5.8.2-2_mips.deb
> -rw-r--r--1 root root 1002 Nov 18 22:31 
> libperl5.8_5.8.2-2_mips.deb
[...]

Can somebody please give it a kick or an upload? Afaict Perl's testing
migration is only blocked by missing builds for mips and mipsel.
 cu andreas

Re: MIPS port backlog, autobuilder machines and some arrogance

[Ingo Juergensmann]

On Wed, Nov 19, 2003 at 02:32:53PM +1100, Martin Michlmayr - Debian Project 
Leader wrote:

> And since we're on topic: people interested in SGI hardware (for
> example to work on debian-installer) in the USA, please get in contact
> with me.

Hmm, regarding to Goswin Brederlow debian-installer build fine yesterday on
our SGI and I'll test it today on my private Indy. 
Perl is building fine as well now on mips, although it is marked as
not-for-us. See
http://m68k.bluespice.org/cgi/package_status?mips_pkg=perl&searchtype=go

paco:/home/ij# ls -l *.deb
-rw-r--r--1 root root35172 Nov 18 22:24 
libcgi-fast-perl_5.8.2-2_all.deb
-rw-r--r--1 root root   651210 Nov 18 22:32 
libperl-dev_5.8.2-2_mips.deb
-rw-r--r--1 root root 1002 Nov 18 22:31 
libperl5.8_5.8.2-2_mips.deb
-rw-r--r--1 root root   766386 Nov 18 22:27 
perl-base_5.8.2-2_mips.deb
-rw-r--r--1 root root  4320256 Nov 18 22:31 
perl-debug_5.8.2-2_mips.deb
-rw-r--r--1 root root  5901334 Nov 18 22:25 perl-doc_5.8.2-2_all.deb
-rw-r--r--1 root root  2158282 Nov 18 22:26 
perl-modules_5.8.2-2_all.deb
-rw-r--r--1 root root32032 Nov 18 22:31 
perl-suid_5.8.2-2_mips.deb
-rw-r--r--1 root root  3325030 Nov 18 22:38 perl_5.8.2-2_mips.deb

-- 
Ciao...  // 
  Ingo \X/

Re: MIPS port backlog, autobuilder machines and some arrogance

[Martin Michlmayr - Debian Project Leader]

* Ingo Juergensmann <[EMAIL PROTECTED]> [2003-11-16 15:40]:
> > Yes, a fairly powerful machine has recently been donated to Debian and
> > we're currently working out where to host it.
> 
> Where is it located?

In the States; not really worth shipping to Germany.  However, I'll
see whether we can find some nice systems in Germany as well.

And since we're on topic: people interested in SGI hardware (for
example to work on debian-installer) in the USA, please get in contact
with me.

-- 
Martin Michlmayr
[EMAIL PROTECTED]

Re: MIPS port backlog, autobuilder machines and some arrogance

[Martin Michlmayr - Debian Project Leader]

* Anthony Towns  [2003-11-17 12:48]:
> On Sun, Nov 16, 2003 at 03:26:51PM +0100, Josip Rodin wrote:
> > Since you're posting that as the DPL you're asking for the following
> > reply. Sorry :)
[...]

I'm well aware that redundancy has several benefits.  Specifically
about the hardware donations manager: I have not received any
complaints in the past, and he properly announced going on vacation
(which are, btw, not real holidays anyway - he's performing some
important Debian duties).

In any case, there was another developer (I think Robert McQueen) who
indicated interest with helping out with hardware donations.  Mako and
he are working on a system which will allow them to coordinate this
task much better; that system will probably also make it easy for
developers to see which offers we currently have.  I will see whether
he can be subscribed to [EMAIL PROTECTED] as a fallback in
the meantime.

-- 
Martin Michlmayr
[EMAIL PROTECTED]

Re: MIPS port backlog, autobuilder machines and some arrogance

[Goswin von Brederlow]

[EMAIL PROTECTED] writes:

> In article <[EMAIL PROTECTED]> you wrote:
> >> Not quite. Ingo is volunteering to provide a MIPS buildd, but since he's
> >> not a Debian Developer, he can't handle its logs. Thus, he's asked me,
> > Its intresting to note that Debian trusts several NMs and normal users
> > to host and maintain their buildds (giving them access to silently
> > backdoor every deb thats build there) but not enough to make them DDs.
> 
> Hmmm, well, to become a DD I would have to apply as a DD, right? And that´s
> what I´m trying to avoid for some certain reasons. 
> Regarding the backdoor issue... you have not the guarantee that even
> certified DDs don´t do this, as well as I don´t have a guarantee that the
> developers that have an account on that machine don´t do something harmfull
> with that... ;))

DDs have to sign and upload a package with a backdoor.

On the buildd I can install a gcc or other tool that will silently add
a backdoor to anything getting compiled and the buildd admin will sign
and upload the package for me.

Much more anonymous.

MfG
Goswin

Re: MIPS port backlog, autobuilder machines and some arrogance

[Josip Rodin]

On Tue, Nov 18, 2003 at 03:32:20PM +1000, Anthony Towns wrote:
> > > > It also means that, if it were easy to add some redundancy,
> > > > it would already have happened. Which in turn means that it's hard.
> > > Again, read what I wrote, not what you imagine I wrote. Difficult isn't
> > > the same as impossible, and hard isn't the same as too hard.
> > So, basically what you're saying that it's hard, and that nobody should be
> > allowed to comment on it because the already delegated people are, what?
> > Perfect? Self-sufficient? Incapable of changing their ways?
> 
> No, I'm saying that nobody who's incapable of assisting with solving the
> problem should be expounding on it. You're welcome to do whatever you
> want on your own time, of course, but if you're going to start accusing
> the DPL, or the buildd maintainers, or anyone else of not doing their
> job on these lists, then you'd better have made absolutely sure you've
> got the knowledge and the experience to back that sort of claim up,
> and that you're able to demonstrate that at all times.

Note that I'm not talking about the buildd issue, but only about the
[EMAIL PROTECTED] thing. I believed that my quoting in the original
indicated that; if not, I'm sorry.

And to answer, yes, I believe I can back up what I said on the topic.
Perhaps there are some intricate details in the system of how donations
are processed, but I rather doubt that it isn't something that a developer
like myself could handle, provided one is diligent enough. (Please don't
interpret this as disrespecting the HD post or Mako -- I certainly don't
indend that.)

> > > BTW, I can't see where I did anything of the sort. I said your post
> > > contributed nothing to the discussion, was unhelpful and distracting and
> > > wrong, and, as such, said that you hadn't contributed anything other
> > > than trite cliches.
> > I don't know about you, but I take it as an insult when someone accuses me
> > of not knowing anything about something[1] and tells me to shut up.
> 
> Again, I never accused you of not knowing anything about this. I said
> that your post didn't demonstrate any knowledge -- "more redundancy is
> good" isn't any more helpful than "too many cooks spoil the broth", or
> "a bird in the hand is worth two in the bush".
> 
> All those things are true, and can be a useful starting point for
> thinking about problems that show up; but they're a starting point only,
> and mindlessly repeating them at people who are already well aware of
> the cliches isn't helpful.

I'm not mindlessly repeating it, nor am I convinced that the people are
already well aware of them (most probably aware of the principle, but
obviously not in the specific case). I don't remember but one instance when
someone said that the hardware donations delegate was unavailable, and the
issue wasn't discussed further or in the proper forum -- and even that
memory is vague so it might have been even less noticeable. To repeat what
I said before, I think that the best, and probably the most logical,
explanation is that the DPL and the delegate simply haven't remembered or
had the chance to consider expanding the hardware donations team -- my
"expounding" on the "cliche" was supposed to be a simple, clear reminder.

> > And there you again. You seem rather inclined to judge other people's
> > competence based on, well, I've no idea on what do you base these claims on.
> 
> Well, an obvious guess would be the posts you've just made. You know,
> the ones I was criticising as being trite and uninformative, while
> pretending at being of profound importance?

Stop accusing me of reading too much into what you wrote when you seem to
have found a pretension at being of profound importance in a comment that
I explicitely marked as obvious and unassuming.

-- 
 2. That which causes joy or happiness.

Re: MIPS port backlog, autobuilder machines and some arrogance

[Anthony Towns]

On Mon, Nov 17, 2003 at 06:25:06PM +0100, Josip Rodin wrote:
> On Tue, Nov 18, 2003 at 02:10:54AM +1000, Anthony Towns wrote:
> > > It also means that, if it were easy to add some redundancy,
> > > it would already have happened. Which in turn means that it's hard.
> > Again, read what I wrote, not what you imagine I wrote. Difficult isn't
> > the same as impossible, and hard isn't the same as too hard.
> So, basically what you're saying that it's hard, and that nobody should be
> allowed to comment on it because the already delegated people are, what?
> Perfect? Self-sufficient? Incapable of changing their ways?

No, I'm saying that nobody who's incapable of assisting with solving the
problem should be expounding on it. You're welcome to do whatever you
want on your own time, of course, but if you're going to start accusing
the DPL, or the buildd maintainers, or anyone else of not doing their
job on these lists, then you'd better have made absolutely sure you've
got the knowledge and the experience to back that sort of claim up,
and that you're able to demonstrate that at all times.

Anything else is both insulting (your post indicates you think Martin
is so stupid that he doesn't know the benefits of redundancy, or that
they're worth considering in this case), and a waste of time and energy
(cliches don't solve hard problems any better than wishful thinking does;
but a good insult can cause all sorts of trouble).

> > BTW, I can't see where I did anything of the sort. I said your post
> > contributed nothing to the discussion, was unhelpful and distracting and
> > wrong, and, as such, said that you hadn't contributed anything other
> > than trite cliches.
> I don't know about you, but I take it as an insult when someone accuses me
> of not knowing anything about something[1] and tells me to shut up.

Again, I never accused you of not knowing anything about this. I said
that your post didn't demonstrate any knowledge -- "more redundancy is
good" isn't any more helpful than "too many cooks spoil the broth", or
"a bird in the hand is worth two in the bush".

All those things are true, and can be a useful starting point for
thinking about problems that show up; but they're a starting point only,
and mindlessly repeating them at people who are already well aware of
the cliches isn't helpful. The problem is finding the right balance,
and that's hard, and requires deep knowledge and experience with the
particular details of the problem. As far as mips autobuilding goes,
indeed as far as _any_ Debian autobuilding goes, I'll take Ryan's opinion
over just about everyone else's, including my own.

> And there you again. You seem rather inclined to judge other people's
> competence based on, well, I've no idea on what do you base these claims on.

Well, an obvious guess would be the posts you've just made. You know,
the ones I was criticising as being trite and uninformative, while
pretending at being of profound importance?

Cheers,
aj

-- 
Anthony Towns <[EMAIL PROTECTED]> 
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda


pgpJAMGrqhgKW.pgp
Description: PGP signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Branden Robinson]

On Mon, Nov 17, 2003 at 12:26:31PM +0100, Wouter Verhelst wrote:
> Op ma 17-11-2003, om 09:58 schreef Anthony Towns:
> > On Mon, Nov 17, 2003 at 01:17:36AM -0500, Nathanael Nerode wrote:
> > > 4) People are volunteering to administer MIPS buildds.
> > 
> > From what I've seen people are volunteering to *provide* MIPS buildds,
> > as long as someone else administers them.
[...]
> However, In reply to a mail I sent to [EMAIL PROTECTED], requesting
> access to the mips wanna-build database and to add the machine to the
> incoming.d.o ACL, Ryan said he'd prefer to remain the only person in
> charge of the mips buildd.

Please.  We can't have actual facts getting in the way of efforts to
close ranks and circle the wagons.

-- 
G. Branden Robinson|Lowery's Law:
Debian GNU/Linux   |If it jams -- force it.  If it
[EMAIL PROTECTED] |breaks, it needed replacing anyway.
http://people.debian.org/~branden/ |


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Josip Rodin]

On Tue, Nov 18, 2003 at 02:10:54AM +1000, Anthony Towns wrote:
> > It also means that, if it were easy to add some redundancy,
> > it would already have happened. Which in turn means that it's hard.
> 
> Again, read what I wrote, not what you imagine I wrote. Difficult isn't
> the same as impossible, and hard isn't the same as too hard.

So, basically what you're saying that it's hard, and that nobody should be
allowed to comment on it because the already delegated people are, what?
Perfect? Self-sufficient? Incapable of changing their ways?

> BTW, I can't see where I did anything of the sort. I said your post
> contributed nothing to the discussion, was unhelpful and distracting and
> wrong, and, as such, said that you hadn't contributed anything other
> than trite cliches.

I don't know about you, but I take it as an insult when someone accuses me
of not knowing anything about something[1] and tells me to shut up.

> you don't have the competence to actually work on.

And there you again. You seem rather inclined to judge other people's
competence based on, well, I've no idea on what do you base these claims on.
Plain old arrogance?

[1] which is false in the case of organizing delegated positions. I have
been doing work on various de jure delegated positions for years now. We can
very well discuss my involvement, performance and any other issue related to
any of them, but it surprises me to no end that someone would try to
entirely nullify the experience I have with this.

-- 
 2. That which causes joy or happiness.

Re: MIPS port backlog, autobuilder machines and some arrogance

[Anthony Towns]

On Mon, Nov 17, 2003 at 12:48:00PM +1000, Anthony Towns wrote:
> It also means that, if it were easy to add some redundancy,
> it would already have happened. Which in turn means that it's hard.

On Mon, Nov 17, 2003 at 03:00:44PM +0100, Josip Rodin wrote:
> > > This NIH attitude is really laughable. 
> > NIH usually stands for "Not Invented Here", meaning someone presuming
> > other people are wrong, and that only ones own ideas are right. You'll
> > note, though, that what I said was that your claim was *trivially
> > obvious*, which is quite a distance from wrong.
> I was responding to your saying that it was "too hard", but you conveniently
> removed that part of the quote.

Again, read what I wrote, not what you imagine I wrote. Difficult isn't
the same as impossible, and hard isn't the same as too hard.

> I don't believe in such a defeatist attitude because the relevant people

Good for you. What makes you think anyone else does? Obviously you do --
your so inclined towards that belief that you're reading things into
what people write that just aren't there.

> (That and an ad hominem attack as an added bonus.)

You know, people love claiming they've had an "ad hominem attack" as
though it makes them some sort of martyr to the cause. It doesn't. An
"ad hominem" fallacy is when you say "you're an idiot, therefore you're
wrong". Saying "you're wrong, therefore you're an idiot" is just a
regular insult.

BTW, I can't see where I did anything of the sort. I said your post
contributed nothing to the discussion, was unhelpful and distracting and
wrong, and, as such, said that you hadn't contributed anything other
than trite cliches. As opposed to saying things like "I see how you
might have a vested interest in trying to defend the acts of the DPL,
given that there've been cases where lack of redundancy among the release
managers caused some difficulties." And hey, you even swore!

So please excuse me if I think you're acting like a pompous idiot, and
wasting everyone's time pontificating on things that you don't have the
competence to actually work on.

HTH, HAND.

Cheers,
aj

-- 
Anthony Towns <[EMAIL PROTECTED]> 
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda


pgpc8wITOPUmI.pgp
Description: PGP signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Josip Rodin]

On Mon, Nov 17, 2003 at 11:10:53PM +1000, Anthony Towns wrote:
> > > > It's been proven plenty of times that whenever we have task depend
> > > > on a single person doing it, the lack of redundancy comes back and
> > > > bites us in the ass whenever there's the slightest bit of a problem.
> > > Why do you think that contributes _anything_ to the discussion? It's not
> > > remotely insightful, rather it's trivially obvious. 
> > This NIH attitude is really laughable. 
> 
> NIH usually stands for "Not Invented Here", meaning someone presuming
> other people are wrong, and that only ones own ideas are right. You'll
> note, though, that what I said was that your claim was *trivially
> obvious*, which is quite a distance from wrong.

I was responding to your saying that it was "too hard", but you conveniently
removed that part of the quote.

I don't believe in such a defeatist attitude because the relevant people
(delegate and/or the leader) haven't provided any evidence to support that
claim, and my past experience with similar tasks and with task management
in general in Debian makes me think quite the contrary.

> > I didn't, however, expect that you'll actually try to sell this kind of
> > bullshit as an actual argument that we need to trust the DPL who is
> > supposedly asserting that things simply had to be done the wrong way.
> 
> No, I'm asserting that people who don't understand what's going on
> well enough to contribute anything other than trite cliches shouldn't
> contribute anything. If you can contribute something more valuable than
> trite cliches, please do; but so far, in this thread, you simply haven't.

Whereas all you've provided is an assertion that nobody knows anything about
the topic except the people that happen to be directly involved in it right
now, which is patently unsupported. (That and an ad hominem attack as an
added bonus.)

-- 
 2. That which causes joy or happiness.

Re: MIPS port backlog, autobuilder machines and some arrogance

[ij]

In article <[EMAIL PROTECTED]> you wrote:
>> Not quite. Ingo is volunteering to provide a MIPS buildd, but since he's
>> not a Debian Developer, he can't handle its logs. Thus, he's asked me,
> Its intresting to note that Debian trusts several NMs and normal users
> to host and maintain their buildds (giving them access to silently
> backdoor every deb thats build there) but not enough to make them DDs.

Hmmm, well, to become a DD I would have to apply as a DD, right? And that´s
what I´m trying to avoid for some certain reasons. 
Regarding the backdoor issue... you have not the guarantee that even
certified DDs don´t do this, as well as I don´t have a guarantee that the
developers that have an account on that machine don´t do something harmfull
with that... ;))

-- 
Ciao...  // 
  Ingo \X/

Re: MIPS port backlog, autobuilder machines and some arrogance

[Anthony Towns]

On Mon, Nov 17, 2003 at 12:37:23PM +0100, Josip Rodin wrote:
> On Mon, Nov 17, 2003 at 12:48:00PM +1000, Anthony Towns wrote:
> > > It's been proven plenty of times that whenever we have task depend on a
> > > single person doing it, the lack of redundancy comes back and bites us in
> > > the ass whenever there's the slightest bit of a problem.
> > Why do you think that contributes _anything_ to the discussion? It's not
> > remotely insightful, rather it's trivially obvious. 
> This NIH attitude is really laughable. 

NIH usually stands for "Not Invented Here", meaning someone presuming
other people are wrong, and that only ones own ideas are right. You'll
note, though, that what I said was that your claim was *trivially
obvious*, which is quite a distance from wrong.

> I see how you might have a vested
> interest in trying to defend the acts of the DPL, given that there've been
> cases where lack of redundancy among the release managers caused some
> difficulties. 

You might like to try reading what I write, rather than assuming that
I'm disagreeing out of some sort of paranoid fear for my job.

You might also like to note the difference between observing a lack of
redundancy, and doing something about it.

> I didn't, however, expect that you'll actually try to sell
> this kind of bullshit as an actual argument that we need to trust the DPL
> who is supposedly asserting that things simply had to be done the wrong way.

No, I'm asserting that people who don't understand what's going on
well enough to contribute anything other than trite cliches shouldn't
contribute anything. If you can contribute something more valuable than
trite cliches, please do; but so far, in this thread, you simply haven't.

Cheers,
aj

-- 
Anthony Towns <[EMAIL PROTECTED]> 
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda


pgpWJ1sQK5TsM.pgp
Description: PGP signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Goswin von Brederlow]

Wouter Verhelst <[EMAIL PROTECTED]> writes:

> Op ma 17-11-2003, om 09:58 schreef Anthony Towns:
> > On Mon, Nov 17, 2003 at 01:17:36AM -0500, Nathanael Nerode wrote:
> > > 4) People are volunteering to administer MIPS buildds.
> > 
> > From what I've seen people are volunteering to *provide* MIPS buildds,
> > as long as someone else administers them.
> 
> Not quite. Ingo is volunteering to provide a MIPS buildd, but since he's
> not a Debian Developer, he can't handle its logs. Thus, he's asked me,

Its intresting to note that Debian trusts several NMs and normal users
to host and maintain their buildds (giving them access to silently
backdoor every deb thats build there) but not enough to make them DDs.

MfG
Goswin

Re: MIPS port backlog, autobuilder machines and some arrogance

[Josip Rodin]

On Mon, Nov 17, 2003 at 12:48:00PM +1000, Anthony Towns wrote:
> > Since you're posting that as the DPL you're asking for the following
> > reply. Sorry :)
> > 
> > It's been proven plenty of times that whenever we have task depend on a
> > single person doing it, the lack of redundancy comes back and bites us in
> > the ass whenever there's the slightest bit of a problem.
> 
> Why do you think that contributes _anything_ to the discussion? It's not
> remotely insightful, rather it's trivially obvious. What that means is
> that Martin already knows it and understands it, and doesn't need to
> be told. It also means that, if it were easy to add some redundancy,
> it would already have happened. Which in turn means that it's hard.

This NIH attitude is really laughable. I see how you might have a vested
interest in trying to defend the acts of the DPL, given that there've been
cases where lack of redundancy among the release managers caused some
difficulties. I didn't, however, expect that you'll actually try to sell
this kind of bullshit as an actual argument that we need to trust the DPL
who is supposedly asserting that things simply had to be done the wrong way.
Even in jokes, the Debian leader isn't ever supposed to be herding sheep...

I like to believe that in the naming of a donations delegate, the DPL
intended the team to evolve as the first delegate gradually got help from
others. Would have been nice to have been done right from the start, but
oh well.

-- 
 2. That which causes joy or happiness.

Re: MIPS port backlog, autobuilder machines and some arrogance

[Wouter Verhelst]

Op ma 17-11-2003, om 09:58 schreef Anthony Towns:
> On Mon, Nov 17, 2003 at 01:17:36AM -0500, Nathanael Nerode wrote:
> > 4) People are volunteering to administer MIPS buildds.
> 
> From what I've seen people are volunteering to *provide* MIPS buildds,
> as long as someone else administers them.

Not quite. Ingo is volunteering to provide a MIPS buildd, but since he's
not a Debian Developer, he can't handle its logs. Thus, he's asked me,
the person currently handling the logs of 'arrakis', his m68k Amiga
which has been running as a buildd since early 2001, whether I would be
willing to handle his MIPS buildd. I have no problem doing so; indeed,
I'd love to expand my work outside of m68k. Since I've been handling
buildd logs since halfway 2001, I'd say I'm qualified.

However, In reply to a mail I sent to [EMAIL PROTECTED], requesting
access to the mips wanna-build database and to add the machine to the
incoming.d.o ACL, Ryan said he'd prefer to remain the only person in
charge of the mips buildd.

>  Running a buildd for Debian
> requires more knowledge than just booting a machine, doing an install,
> pointing wanna-build at auric and crossing your fingers.

Sure; that's why nobody's suggesting that.

-- 
Wouter Verhelst
Debian GNU/Linux -- http://www.debian.org
Nederlandstalige Linux-documentatie -- http://nl.linux.org
If you're running Microsoft Windows, either scan your computer on
viruses, or stop wasting my bandwith and remove me from your
addressbook. *now*.


signature.asc
Description: Dit berichtdeel is digitaal ondertekend

Re: MIPS port backlog, autobuilder machines and some arrogance

[Anthony Towns]

On Mon, Nov 17, 2003 at 01:17:36AM -0500, Nathanael Nerode wrote:
> 4) People are volunteering to administer MIPS buildds.

From what I've seen people are volunteering to *provide* MIPS buildds,
as long as someone else administers them. Running a buildd for Debian
requires more knowledge than just booting a machine, doing an install,
pointing wanna-build at auric and crossing your fingers.

Cheers,
aj

-- 
Anthony Towns <[EMAIL PROTECTED]> 
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda


pgphWsUYNvJA3.pgp
Description: PGP signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Marcelo E. Magallon]

On Mon, Nov 17, 2003 at 02:46:07PM +1100, Martin Michlmayr - Debian Project 
Leader wrote:

 > Actually, Linux has been ported to all of the CPUs and systems I had
 > in mind; in fact, ported by the vendor since there's demand for Linux.
 > The 1 GHz dual-cores I mentioned are one from Broadcom (which drow
 > also mentioned) and another one developed by PMC-Sierra.
 > 
 > See http://www.broadcom.com/products/product.php?product_id=BCM1250
 > and http://www.momenco.com/products/jag-atx.html

 Ah... I suspected that :-)  "_very_ powerful" is a matter of
 perspective.  While the newer MIPS processors are indeed very fine
 pieces of hardware, these machines are targeted towards very specific
 markets where requirements are slightly different from those in the HPC
 community.  Sure, noone will stop you from getting decent connectivity
 on these things and putting a bunch of them in a cluster configuration
 (and considering the historical record of the MIPS architecture in the
 floating point arena, this is probably an attractive proposition... if
 the price tag was a bit lower)

 _I_ was thinking of something with a "somewhat" larger power
 consumption.  One can only hope SGI will survive long enough...

 Thanks for the info to both of you guys,

 Marcelo,
 wondering what do the MIPS folk do with Debian in the real world...

Re: MIPS port backlog, autobuilder machines and some arrogance

[Nathanael Nerode]

Josip Robin wrote:
It's been proven plenty of times that whenever we have task depend on a
single person doing it, the lack of redundancy comes back and bites us in
the ass whenever there's the slightest bit of a problem.
Anthony Towns then wrote:
Why do you think that contributes _anything_ to the discussion? It's not
remotely insightful, rather it's trivially obvious. What that means is
that Martin already knows it and understands it, and doesn't need to
be told. It also means that, if it were easy to add some redundancy,
it would already have happened. Which in turn means that it's hard.
No.  You're saying "Because nobody has done it, it must be hard," which 
just isn't the way the world works; it makes so many wrong assumptions 
it's not even worth enumerating them.

The specific case at hand is actually a counterexample to your 
argument.  Let me list the facts I've heard on
this list:
1) There is a single MIPS buildd admin.
2) He is the point of failure, for whatever reason.
3) Other architectures successfully have multiple buildd admins.
4) People are volunteering to administer MIPS buildds.
5) Yet, there is still a single MIPS buildd admin.

Don't overlook the degree to which people can overlook the "trivially 
obvious".

Re: MIPS port backlog, autobuilder machines and some arrogance

[Martin Michlmayr - Debian Project Leader]

* Marcelo E. Magallon <[EMAIL PROTECTED]> [2003-11-16 12:10]:
>  > [ Responding to this since I've been BCCed. ]
>  Oh, nice.  If hadn't been Bcc'ed you wouldn't respond?

I didn't say that; I simply stressed the importance of (B)CCing
[EMAIL PROTECTED] if you want to ensure getting a reply from the DPL.  I
don't read all mailing lists or threads; but I happend to read this
one anyway.

> There are some newer "_very_ powerful" machines with MIPS processors,
> but AFAIK noone has actually ported Linux to them.

Actually, Linux has been ported to all of the CPUs and systems I had
in mind; in fact, ported by the vendor since there's demand for Linux.
The 1 GHz dual-cores I mentioned are one from Broadcom (which drow
also mentioned) and another one developed by PMC-Sierra.

See http://www.broadcom.com/products/product.php?product_id=BCM1250
and http://www.momenco.com/products/jag-atx.html

Wind River has a good table listing most newer mips boards which are
available:
http://www.windriver.com/products/bsp_web/bsp_architecture.html?architecture=MIPS
Also, see the press release from MIPS to see what some companies have
been working on:
http://www.mips.com/content/PressRoom/PressReleases/2003-04-23

-- 
Martin Michlmayr
[EMAIL PROTECTED]

Re: MIPS port backlog, autobuilder machines and some arrogance

[Anthony Towns]

On Sun, Nov 16, 2003 at 03:26:51PM +0100, Josip Rodin wrote:
> Since you're posting that as the DPL you're asking for the following
> reply. Sorry :)
> 
> It's been proven plenty of times that whenever we have task depend on a
> single person doing it, the lack of redundancy comes back and bites us in
> the ass whenever there's the slightest bit of a problem.

Why do you think that contributes _anything_ to the discussion? It's not
remotely insightful, rather it's trivially obvious. What that means is
that Martin already knows it and understands it, and doesn't need to
be told. It also means that, if it were easy to add some redundancy,
it would already have happened. Which in turn means that it's hard.

If you want to help these things, first you have to recognise the _actual_
problems, then see if you can help solve them.

Doing the reverse -- thinking of problems you could address, then
hoping they exist, like say, imagining Martin doesn't see the benefits
in redundancy, and solving that by telling him -- is both unhelpful
and distracting.

(And, actually it's wrong, it bites us when the person responsible for
the job disappears or is busy doing other things -- "the slightest bit
of a problem" is usually handled so well you don't even notice. A little
less hyperbole would be nice.)

Cheers,
aj

-- 
Anthony Towns <[EMAIL PROTECTED]> 
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda


pgp4jm5wBwjFM.pgp
Description: PGP signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Goswin von Brederlow]

Daniel Jacobowitz <[EMAIL PROTECTED]> writes:

> There's one on my desk.  It's an embedded board produced by Broadcom,
> complete with dual cores, IDE, and dual gigabit ethernet.  MIPS, Inc.
> has a similar board; Debian has one of these somewhere, IIRC.  These
> little fellows have more computing power than our PPC build daemon.
> 
> The machines you are likely to be thinking of are things like the SGI
> R12000; the conventional workstation-type MIPS platforms have much
> worse support than the embedded ones, but the embedded ones pack a
> serious punch nowadays.

I want one too. If they would just come without the 5000$ development
support.

MfG
Goswin

Re: MIPS port backlog, autobuilder machines and some arrogance

[Daniel Jacobowitz]

On Sun, Nov 16, 2003 at 12:10:14PM -0600, Marcelo E. Magallon wrote:
> On Mon, Nov 17, 2003 at 12:53:39AM +1100, Martin Michlmayr - Debian Project 
> Leader wrote:
>  > [ Responding to this since I've been BCCed. ]
> 
>  Oh, nice.  If hadn't been Bcc'ed you wouldn't respond?  That's
>  reasuring, since you seem to be one of the few persons who's got some
>  information on the subject and are somewhat willing to talk.
> 
>  > Also, while we are covering this topic, let me mention that mips is
>  > not a "door stop" architecture.  While most machines which were
>  > _very_ powerful in the past are fairly dated now (e.g. SGI Indy),
>  > mips CPUs are actively being developed.  There are CPUs with 2 cores
>  > and > 1 GHz frequency.
> 
>  Care to give an example of one of this "_very_ powerful" machines,
>  which are in production and where Linux actually runs?  The "_very_
>  powerful" SGI Indy was attractive at its time only because of some of
>  the graphics boards it supported, none of which are supported on Linux.
>  There are some newer "_very_ powerful" machines with MIPS processors,
>  but AFAIK noone has actually ported Linux to them.  There's one "_very_
>  powerful" machine using MIPS processors which supposedly will be made
>  available with Linux on it in the future, yet it's already shadowed by
>  machines using AMD64 processors which happen to be available now and
>  where Linux runs now with support for all the functionality of the
>  hardware.
> 
>  Just curious...

There's one on my desk.  It's an embedded board produced by Broadcom,
complete with dual cores, IDE, and dual gigabit ethernet.  MIPS, Inc.
has a similar board; Debian has one of these somewhere, IIRC.  These
little fellows have more computing power than our PPC build daemon.

The machines you are likely to be thinking of are things like the SGI
R12000; the conventional workstation-type MIPS platforms have much
worse support than the embedded ones, but the embedded ones pack a
serious punch nowadays.

-- 
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer

Re: MIPS port backlog, autobuilder machines and some arrogance

[Marcelo E. Magallon]

On Mon, Nov 17, 2003 at 12:53:39AM +1100, Martin Michlmayr - Debian Project 
Leader wrote:
 > [ Responding to this since I've been BCCed. ]

 Oh, nice.  If hadn't been Bcc'ed you wouldn't respond?  That's
 reasuring, since you seem to be one of the few persons who's got some
 information on the subject and are somewhat willing to talk.

 > Also, while we are covering this topic, let me mention that mips is
 > not a "door stop" architecture.  While most machines which were
 > _very_ powerful in the past are fairly dated now (e.g. SGI Indy),
 > mips CPUs are actively being developed.  There are CPUs with 2 cores
 > and > 1 GHz frequency.

 Care to give an example of one of this "_very_ powerful" machines,
 which are in production and where Linux actually runs?  The "_very_
 powerful" SGI Indy was attractive at its time only because of some of
 the graphics boards it supported, none of which are supported on Linux.
 There are some newer "_very_ powerful" machines with MIPS processors,
 but AFAIK noone has actually ported Linux to them.  There's one "_very_
 powerful" machine using MIPS processors which supposedly will be made
 available with Linux on it in the future, yet it's already shadowed by
 machines using AMD64 processors which happen to be available now and
 where Linux runs now with support for all the functionality of the
 hardware.

 Just curious...

 Marcelo

Re: MIPS port backlog, autobuilder machines and some arrogance

[John Hasler]

Josip Rodin writes:
> It's been proven plenty of times that whenever we have task depend on a
> single person doing it, the lack of redundancy comes back and bites us in
> the ass whenever there's the slightest bit of a problem.

Example: I recently had an opportunity to arrange the donation of five
RS/6000s.  I emailed the donations manager and got no reply (not even a
vacation message).  The donor needed to get rid of the machines ASAP so
they went elsewhere.
-- 
John Hasler
[EMAIL PROTECTED] (John Hasler)
Dancing Horse Hill
Elmwood, WI

Re: MIPS port backlog, autobuilder machines and some arrogance

[Ingo Juergensmann]

On Mon, Nov 17, 2003 at 12:53:39AM +1100, Martin Michlmayr - Debian Project 
Leader wrote:
> [ Responding to this since I've been BCCed. ]
> * Ingo Juergensmann <[EMAIL PROTECTED]> [2003-11-14 11:34]:
> > I´ve been told that this is caused by some toolchain and kernel
> > problems on mips that make the buildds running slow.
> I talked to Ryan about the mips port about two weeks ago; I'll try to
> summarize the information he gave me.  The kernel problem you refer to

Great. As already stated, I've contacted him as well last week without any
response to the MIPS situation neither to http://buildd.debian.org linking
or even hosting the buildd stats for all archs currently hosted on
http://m68k.bluespice.org.
No response leads to the assumption that the person is unresponsive
(vacation, busy, ignorant, ill, MIA, ...).   

> is that the kernel does not use the secondary cache of the chips.

Yes, and that's another reason why more machines would be more helpful,
IMHO.

> Another problem is that one machine (hosted by Flo in Germany) only
> has 64 MB RAM.  Apparently, Flo has more RAM, though; otherwise, I

Arrakis (the shutdown m68k buildd) has only 64 MB as well and is performing
comparably great. 
The offered mips machine has 96 MB - but there was no conversation about how
well equipped the machine is or not. The offer was just dropped down by
obscure reasons.

> have recently been given some RAM which I'll send him once I'll be
> back in Europe (in December).

If you still have that RAM, there is just another Indigo2 we can offer here
in Rostock. ;-)
 
> The major problem why mips is not keeping up, however, is that the
> hard drive of a build machine in the States died recently.

Well, same happened to m68k port over the last weeks and that's the reason
why I believe that having more machines is always better, because there'll
always fail one machine or another for some reasons, so it's good to have
backup machines.

> Unfortunately, this machine is much faster than the other machines in
> use.  I am trying to get a new hard drive for this machine, but
> unfortunately the hardware donations manager is currently on vacation
> and hasn't responded to my mail yet; we might just buy one if he
> doesn't respond soon.  Alternatively, in anyone is reading this who
> can spare a SCSI disk (in the States, for an Indy, see
> http://sgistuff.g-lenerz.de/machines/indy.html), please get in
> contact.  This would help to improve the situation drastically.

Disks donations are always appreciated. ;)
 
> > Everything went well with that machine - until we directed the
> > request to debian-admin to get wanna-build access for mips. The
> > request was rejected with the following reasons (to my knowledge):
> > - another machine is in the works
> > - we don´t need your machine
> Yes, a fairly powerful machine has recently been donated to Debian and
> we're currently working out where to host it.

Where is it located?

>  Also, I have been in
> contact with SGI to get a very powerful system which should solve this
> problem for a long time.  However, I am still working on this.

It would be nice to have SGI help out for porting the kernel to the newer
MIPS CPUs such as R12000. We have an Octane R12000 standing there that could
be used... well, you know it... ;))
 
> I also offered Ryan to get a temporary system online, but he told me
> not to bother unless it's a very high performance system.  IIUC, it
> takes some time to get it set up as a buildd, etc.

Hmmm... I don't know how long Ryan needs to setup a buildd, but for m68k we
have already setup enough buildds as well and everything is there and just
needs a "Go for it!" and the access to wanna-build db and IP added for
access to incoming. This could be done within some hours. 
Debian is running on that machine, mail is running, IP is static, buildd and
sbuild are installed 

> In any case, your offer is welcome, and I think we can make use of it,
> albeit in a different way.  You could make the box available to
> various Debian people.  For example, the XFree86 maintainer is looking
> for someone to try building the current version in experimental on
> mips.  That would be a great way your machine could be used.

XFree needs a little too more disk space than the machine currently has, but
there is already a solution in progress. Maybe when someone wants to donate
another disk, it would be helpful as well... ;) 

> In any case, the situation is known and solutions are being worked on.
> Help is definitely appreciated. 

That's not the impression I got. 
The answer of "we don't want your machine" was rather harsh and arrogant in
my eyes, especially the reasoning that I'm not experienced enough on MIPS to
run a buildd is a rather stupid one.
I'm complaining about the way this was communicated, that the pure fact that
the machine isn't needed as a buildd itself. Some people should really work
on their social skills, I guess... :-//

> Also, while we are covering this
> top

Re: MIPS port backlog, autobuilder machines and some arrogance

[Josip Rodin]

On Mon, Nov 17, 2003 at 12:53:39AM +1100, Martin Michlmayr - Debian Project 
Leader wrote:
> I am trying to get a new hard drive for this machine, but unfortunately
> the hardware donations manager is currently on vacation and hasn't
> responded to my mail yet; we might just buy one if he doesn't respond
> soon.

Since you're posting that as the DPL you're asking for the following
reply. Sorry :)

It's been proven plenty of times that whenever we have task depend on a
single person doing it, the lack of redundancy comes back and bites us in
the ass whenever there's the slightest bit of a problem.

-- 
 2. That which causes joy or happiness.

Re: MIPS port backlog, autobuilder machines and some arrogance

[Martin Michlmayr - Debian Project Leader]

[ Responding to this since I've been BCCed. ]

* Ingo Juergensmann <[EMAIL PROTECTED]> [2003-11-14 11:34]:
> I´ve been told that this is caused by some toolchain and kernel
> problems on mips that make the buildds running slow.

I talked to Ryan about the mips port about two weeks ago; I'll try to
summarize the information he gave me.  The kernel problem you refer to
is that the kernel does not use the secondary cache of the chips.
Another problem is that one machine (hosted by Flo in Germany) only
has 64 MB RAM.  Apparently, Flo has more RAM, though; otherwise, I
have recently been given some RAM which I'll send him once I'll be
back in Europe (in December).

The major problem why mips is not keeping up, however, is that the
hard drive of a build machine in the States died recently.
Unfortunately, this machine is much faster than the other machines in
use.  I am trying to get a new hard drive for this machine, but
unfortunately the hardware donations manager is currently on vacation
and hasn't responded to my mail yet; we might just buy one if he
doesn't respond soon.  Alternatively, in anyone is reading this who
can spare a SCSI disk (in the States, for an Indy, see
http://sgistuff.g-lenerz.de/machines/indy.html), please get in
contact.  This would help to improve the situation drastically.

> Everything went well with that machine - until we directed the
> request to debian-admin to get wanna-build access for mips. The
> request was rejected with the following reasons (to my knowledge):
> - another machine is in the works
> - we don´t need your machine

Yes, a fairly powerful machine has recently been donated to Debian and
we're currently working out where to host it.  Also, I have been in
contact with SGI to get a very powerful system which should solve this
problem for a long time.  However, I am still working on this.

I also offered Ryan to get a temporary system online, but he told me
not to bother unless it's a very high performance system.  IIUC, it
takes some time to get it set up as a buildd, etc.

In any case, your offer is welcome, and I think we can make use of it,
albeit in a different way.  You could make the box available to
various Debian people.  For example, the XFree86 maintainer is looking
for someone to try building the current version in experimental on
mips.  That would be a great way your machine could be used.

In any case, the situation is known and solutions are being worked on.
Help is definitely appreciated.  Also, while we are covering this
topic, let me mention that mips is not a "door stop" architecture.
While most machines which were _very_ powerful in the past are fairly
dated now (e.g. SGI Indy), mips CPUs are actively being developed.
There are CPUs with 2 cores and > 1 GHz frequency.

> As a result and a sort of protest, I´ll stopped my m68k buildd,
> because I don´t know m68k that much to be of any help for this port
> anymore.

A m68k has indicated that your machine is helpful for the m68k port,
so I urge you to reconsider this decision.

-- 
Martin Michlmayr
[EMAIL PROTECTED]

Re: MIPS port backlog, autobuilder machines and some arrogance

[Goswin von Brederlow]

Marco d'Itri <[EMAIL PROTECTED]> writes:

> On Nov 15, Chris Cheney <[EMAIL PROTECTED]> wrote:
> 
>  >Another point is do we really want an arch/port to be maintained by only
>  >one person? IMHO there should be at least two people capable and
>  >actually running the buildds for each arch, possibly more when they are
>  >too busy with other things as appears the case with Ryan.
> I fully agree. If a port has not enough developers to care about it,
> then it should not be considered a candidate for releases.

There is a difference between available/existing and responcible.

In this case developer who want to help out with hardware and time are
put down by one power hungry (no offence ment, don't even know who you
are, exagerated to make a point) DD who is blocking any help offered.

MfG
Goswin

PS: realy no offence ment, I just wish other archs where as friendly and open 
ad the m68k buildd community

Re: MIPS port backlog, autobuilder machines and some arrogance

[Marco d'Itri]

On Nov 15, Chris Cheney <[EMAIL PROTECTED]> wrote:

 >Another point is do we really want an arch/port to be maintained by only
 >one person? IMHO there should be at least two people capable and
 >actually running the buildds for each arch, possibly more when they are
 >too busy with other things as appears the case with Ryan.
I fully agree. If a port has not enough developers to care about it,
then it should not be considered a candidate for releases.

-- 
ciao, |
Marco | [3071 sfP4YzRx/2xIc]


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Ingo Juergensmann]

On Sat, Nov 15, 2003 at 12:54:18AM -0500, Branden Robinson wrote:
> On Fri, Nov 14, 2003 at 11:34:41AM +0100, Ingo Juergensmann wrote:
> > As some of you know the mips port has some problems keeping up.
> Daniel Stone and I have been trying for months to get feedback regarding
> xfree86 > 4.3.0-0pre1v1 on mips, and we are always met with stony
> silence.

Right. He had the chance to say "No, thx, that backlog is just temporarily
and we solved the backlog and having another machine is not needed" in a
polite way for about a week, but nothing happened. 

> > Everything went well with that machine - until we directed the request to
> > debian-admin to get wanna-build access for mips. The request was rejected
> > with the following reasons (to my knowledge):
> > - another machine is in the works
> Which, of course, will never go down, suffer hardware failure, or have
> its hosting site suffer a power outage or fire.

Right. On m68k we have the most buildd machines and guess what: murphy
looked in some weeks and several buildds went away nearly at the same time
giving a backlog of about >220 packages rather quick. 
Therefor we don't say "Oh my god! We don't need so many buildds because more
machines can fail more often!" but more likely "Hey, the number of packages
is constantly growing, the toolchain is very often giving problems and is
getting slower as well (gcc3.3) and there will always be exists problems
that take machines down. So we appreciate to have even more machines and
have a pool of buildd maintainers as well to help each other!"

> > - we don´t need your machine
> And we never will; see above.

On m68k-build there was a little more discussion about the reasons. See the
webarchive for the list. 
Funny thing is, that every responsible people gives another reason for
rejecting our offer, which make me feel that not the machine is the problem
but that those people want to act alone on their own will and decision. 
The reason that were given there was that having multiple buildd maintainer
will make it hard to work together and it's more simple when a single buildd
maintainer manages all of the buildds for one arch.
a) I think that's not true. m68k shows clearly that this can work and even
is a benefit, because I think that m68k is the most responsive and helpful
archs when maintainers direct questions to the (right ;) list. 
b) When it is a problem for a person to work together with others he clearly
should reconsider his involvment in Debian, because the project is a
cooperate effort and not something where helping hands should be excluded
when they want and are able to help. Although I don't think it's violating
the social contract, I don't think that this behaviour is in the sense of
it. 
c) We offered a machine. Maybe it's not the fastest or best equipped right
now but it is usuable. And we didn't insisted of administrating it by our
own, although Wouter wanted to maintain the buildd in order to unburden
Ryan, but when he wanted to take it himself that would be ok as well. 
d) There is a second machine that could be used as buildd, when the new
disks and the memory arrives. All machines would have fixed IP addresses, so
that the machines could even be used as public machines. But the responsible
persons have not asked for either option but simply rejected the idea of
having more machines to help with the backlog without asking for further
information. And that's plain stupid or arrogant. Call it as you want.

> > - you don´t have any knowledge about mips, so the machine wouldn´t be of any
> > help at all.
> It's not like you're an experienced buildd admin, and it's vitally
> important that people who aren't already experts at administrating mips
> buildds not gain that expertise.

It's even more worse! 
I simply cannot be a buildd admin on my own, because I'm not a DD and thus
cannot sign the packages. And to improve the worseness: I understand as much
mips asm as I understand m68k asm: not a single word! 
So, saying I'm not experienced enough on mips also means I cannot run a m68k
buildd, because my knowledge is the same for both archs. ;-)

> > As a result and a sort of protest, I´ll stopped my m68k buildd, because I
> > don´t know m68k that much to be of any help for this port anymore. Therefore
> > my m68k isn´t needed anymore as my offered mips machine isn´t needed for the
> > mips port or the Debian project at all.
> I'm not sure I agree with your decision, but I do think I understand
> your frustration.

Thx.

> Can I ask why it is such a disaster to have an alternate or standby
> buildd for the mips architecture?

IMHO because some people don't want anyone else to play on their playground.

-- 
Ciao...  // 
  Ingo \X/

Re: MIPS port backlog, autobuilder machines and some arrogance

[Chris Cheney]

Another point is do we really want an arch/port to be maintained by only
one person? IMHO there should be at least two people capable and
actually running the buildds for each arch, possibly more when they are
too busy with other things as appears the case with Ryan.

Chris


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Chris Cheney]

On Fri, Nov 14, 2003 at 02:55:09PM +0100, Wouter Verhelst wrote:
> Op vr 14-11-2003, om 11:34 schreef Ingo Juergensmann:
> [...]
> > As a result and a sort of protest, I´ll stopped my m68k buildd, because I
> > don´t know m68k that much to be of any help for this port anymore. Therefore
> > my m68k isn´t needed anymore as my offered mips machine isn´t needed for the
> > mips port or the Debian project at all.
> 
> Ingo,
> 
> I can understand why you're upset, but please do try not to make one
> port suffer for the actions of the people responsible for another port.
> The help arrakis has provided over the years has always been
> appreciated, and will be for as long as you provide the access; it would
> be a shame if this would be discontinued because of a difference in
> opinion you have with Ryan regarding the way autobuilding for the mips
> architecture should be handled.

So why does Ryan have the authority to reject more buildd's when he
obviously is overburdened to the point he doesn't respond to emails
(see Branden Robinson's post as an example). From an untrained eye the
mips buildd's seem to be building packages more or less fine just that
they were lagged by the human behind them. One example is kdelibs which
wasn't built for nearly a month due to what should have been a Dep-Wait
on a fixed libXrender, which was uploaded the same day mips attempted to
build kdelibs.

Chris


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Branden Robinson]

On Fri, Nov 14, 2003 at 11:34:41AM +0100, Ingo Juergensmann wrote:
> As some of you know the mips port has some problems keeping up.

Daniel Stone and I have been trying for months to get feedback regarding
xfree86 > 4.3.0-0pre1v1 on mips, and we are always met with stony
silence.

> Everything went well with that machine - until we directed the request to
> debian-admin to get wanna-build access for mips. The request was rejected
> with the following reasons (to my knowledge):
> - another machine is in the works

Which, of course, will never go down, suffer hardware failure, or have
its hosting site suffer a power outage or fire.

> - we don´t need your machine

And we never will; see above.

> - you don´t have any knowledge about mips, so the machine wouldn´t be of any
> help at all.

It's not like you're an experienced buildd admin, and it's vitally
important that people who aren't already experts at administrating mips
buildds not gain that expertise.

> As a result and a sort of protest, I´ll stopped my m68k buildd, because I
> don´t know m68k that much to be of any help for this port anymore. Therefore
> my m68k isn´t needed anymore as my offered mips machine isn´t needed for the
> mips port or the Debian project at all.

I'm not sure I agree with your decision, but I do think I understand
your frustration.

Can I ask why it is such a disaster to have an alternate or standby
buildd for the mips architecture?

-- 
G. Branden Robinson|Kissing girls is a goodness.  It is
Debian GNU/Linux   |a growing closer.  It beats the
[EMAIL PROTECTED] |hell out of card games.
http://people.debian.org/~branden/ |-- Robert Heinlein


signature.asc
Description: Digital signature

Re: MIPS port backlog, autobuilder machines and some arrogance

[Wouter Verhelst]

Op vr 14-11-2003, om 11:34 schreef Ingo Juergensmann:
[...]
> As a result and a sort of protest, I´ll stopped my m68k buildd, because I
> don´t know m68k that much to be of any help for this port anymore. Therefore
> my m68k isn´t needed anymore as my offered mips machine isn´t needed for the
> mips port or the Debian project at all.

Ingo,

I can understand why you're upset, but please do try not to make one
port suffer for the actions of the people responsible for another port.
The help arrakis has provided over the years has always been
appreciated, and will be for as long as you provide the access; it would
be a shame if this would be discontinued because of a difference in
opinion you have with Ryan regarding the way autobuilding for the mips
architecture should be handled.

Hoping you'll reconsider this,

-- 
Wouter Verhelst
Debian GNU/Linux -- http://www.debian.org
Nederlandstalige Linux-documentatie -- http://nl.linux.org
If you're running Microsoft Windows, either scan your computer on
viruses, or stop wasting my bandwith and remove me from your
addressbook. *now*.


signature.asc
Description: Dit berichtdeel is digitaal ondertekend

Re: MIPS port backlog, autobuilder machines and some arrogance

[Andreas Tille]

On Fri, 14 Nov 2003, Ingo Juergensmann wrote:

> As some of you know the mips port has some problems keeping up.
I guess the relevant reference for this problem is:

 http://buildd.debian.org/stats/graph-week-big.png

If you ask me (as a person without any knowledge about buildd internals)
a further mips machine could do some help here.

Kind regards

  Andreas.

MIPS port backlog, autobuilder machines and some arrogance

[Ingo Juergensmann]

As some of you know the mips port has some problems keeping up.

I´ve been told that this is caused by some toolchain and kernel problems on
mips that make the buildds running slow.

So, when watching http://m68k.bluespice.org/buildd/mips_stats over the past
weeks, it´s obvious that mips is heavily struggling on its backlog. 

On m68k we usually know all of those problems as well and having many
machines always helped a lot when one machine or another failed to work
properly. So, the idea was to help themips port with additional mips
machines that can be used as buildds. 

Everything went well with that machine - until we directed the request to
debian-admin to get wanna-build access for mips. The request was rejected
with the following reasons (to my knowledge):
- another machine is in the works
- we don´t need your machine
- you don´t have any knowledge about mips, so the machine wouldn´t be of any
help at all.

As a result and a sort of protest, I´ll stopped my m68k buildd, because I
don´t know m68k that much to be of any help for this port anymore. Therefore
my m68k isn´t needed anymore as my offered mips machine isn´t needed for the
mips port or the Debian project at all.

-- 
Ciao...  // 
  Ingo \X/