about GNU Hurd

[Stephan Peijnik-3]

On Wed, Jul 25, 2007 at 09:21:30PM +0200, Alfred M. Szmidt wrote:
> [...] and new people joining the project get immensly
> confused on what exactly they should work on since everyone is saying
> `work on what you want'.  It is a whole `no care attitude' which is
> simply destroying the project.

I can only agree to that. I have been thinking about doing some Hurd hacking
in the past, but after seeing that a lot of patches which should have been
commited long ago still have not been applied it seemed like the Hurd is
quite dead.

I am sure there are quite a few people out there who would love to
contribute to the development of the Hurd and have not done so because
it looks as if also patches which are both fine and needed have not been
applied.

Speaking of myself I did not do any hacking on the Hurd because I do not
want to write code that is not going to be applied for whatever reasons.

In my opinion, what the Hurd needs right now, and first of all, is
a maintainer which applies patches and, once that works, some sort of
advertising to attract new hackers.

[Thomas Schwinge-6]

On Wed, Jul 25, 2007 at 09:21:30PM +0200, Alfred M. Szmidt wrote:
>    Registering patches in the glibc bugzilla had been made a
>    requirement by the glibc (release) maintainers to get them into the
>    non-HEAD branches, and since that was to a large share what we've
>    been working on, this seemed the most suitable way to publish them,
>    additionally to having published them on the bug-hurd mailing list.
>
> The Hurd only follows HEAD.

I'm curious: why are you to state what the Hurd does and what not?


 T. Schwinge

[Richard Stallman]

    In my opinion, what the Hurd needs right now, and first of all, is
    a maintainer which applies patches and, once that works, some sort of
    advertising to attract new hackers.

I think you are right.

[Alfred M. Szmidt]

   >    Registering patches in the glibc bugzilla had been made a
   >    requirement by the glibc (release) maintainers to get them into the
   >    non-HEAD branches, and since that was to a large share what we've
   >    been working on, this seemed the most suitable way to publish them,
   >    additionally to having published them on the bug-hurd mailing list.
   >
   > The Hurd only follows HEAD.

   I'm curious: why are you to state what the Hurd does and what not?

'cause uncle Rollie said so.

[Thomas Schwinge-6]

On Thu, Jul 26, 2007 at 08:19:17PM +0200, Alfred M. Szmidt wrote:

>    >    Registering patches in the glibc bugzilla had been made a
>    >    requirement by the glibc (release) maintainers to get them into the
>    >    non-HEAD branches, and since that was to a large share what we've
>    >    been working on, this seemed the most suitable way to publish them,
>    >    additionally to having published them on the bug-hurd mailing list.
>    >
>    > The Hurd only follows HEAD.
>
>    I'm curious: why are you to state what the Hurd does and what not?
>
> 'cause uncle Rollie said so.

Please go on, that's interesting.  You say that ``uncle Rollie'' -- as
you call him -- enabled you to state what the Hurd does and what not?
That's surprising given that you have explicitly been excluded from the
Hurd project with the Hurd maintainers consent.

Save that, ``uncle Rollie'' also was the one to tell me to register
patches we want to get into the glibc non-HEAD branches in the glibc
bugzilla.


 T. Schwinge

[Alfred M. Szmidt]

If you wish to look up what is being followed, please refer to the
libc-alpha, bug-hurd, etc, mailing lists.

[Luis Araujo-2]

Thomas Schwinge wrote:

> On Thu, Jul 26, 2007 at 08:19:17PM +0200, Alfred M. Szmidt wrote:
>>    >    Registering patches in the glibc bugzilla had been made a
>>    >    requirement by the glibc (release) maintainers to get them into the
>>    >    non-HEAD branches, and since that was to a large share what we've
>>    >    been working on, this seemed the most suitable way to publish them,
>>    >    additionally to having published them on the bug-hurd mailing list.
>>    >
>>    > The Hurd only follows HEAD.
>>
>>    I'm curious: why are you to state what the Hurd does and what not?
>>
>> 'cause uncle Rollie said so.
>
> Please go on, that's interesting.  You say that ``uncle Rollie'' -- as
> you call him -- enabled you to state what the Hurd does and what not?
> That's surprising given that you have explicitly been excluded from the
> Hurd project with the Hurd maintainers consent.
>
> Save that, ``uncle Rollie'' also was the one to tell me to register
> patches we want to get into the glibc non-HEAD branches in the glibc
> bugzilla.
>
>
>  T. Schwinge

Or probably, because he is one of the few people who has actually done
some 'real' work on the Hurd around here and he has won that right at
some degree.

Regards,

[Richard Stallman]

    For getting invasive (read: interface changes and the like) changes into
    the Hurd we'd need (a) people to come up with such changes (!) and (b)
    have them attract Thomas Bushnell's or Roland McGrath's or Marcus
    Brinkmann's attention (which often enough is not trivial).

    But I deem it's very difficult to find someone to judge on such issues
    and have that person be a judge with Thomas Bushnell's or Roland
    McGrath's or Marcus Brinkmann's qualities.

    I can't be that person, and neither can -- in my opinion -- be the other
    people who took part in this discussion so far, as everyone of us lacks
    the experience to judge on such things.

I appreciate your modesty.  However, the expertise of those old Hurd
developers does us no good if they don't actually use it.  It would be
better for you to install these changes than for us to wait forever.

Would you be willing to do it?

[Richard Stallman]

    In fact while I am being frank, what is your level of interest in the
    Hurd Richard?  The few times that I have seen you speak or read your
    commentaries, you don't show much of an interest yourself.  You speak of
    Ututo and others when you have legions of folks that you could probably
    drive to the "GNU System".

When we started development of the Hurd, it was absolutely essential.
We needed a free kernel, and there was none.

Since we have Linux, the Hurd is no longer essential.  Making the Hurd
work well would be good for two reasons:

* Because it would be more powerful than Linux.

* Because it would be a GNU kernel.

The unfortunate thing is that making the Hurd really secure
requires Coyotos.  I just asked Marcus Brinkmann how progress
is on Coyotos.

[Claudio Fontana-3]

Hello,

On 7/25/07, Stephan Peijnik <sp@...> wrote:
> In my opinion, what the Hurd needs right now, and first of all, is
> a maintainer which applies patches and, once that works, some sort of
> advertising to attract new hackers.

About the last part, maybe some kind of "bounty" could be set up?
As this is hard work, and maybe not immediately rewarding, maybe
some tangible monetary reward on a per-feature or per-milestone
basis could attract new hackers?

I am not familiar with the Hurd and its development status, and do not
feel able to contribute to its development directly, but I am able to offer a
moderate bounty for a stable, GPLv3, more or less drop-in replacement
of the Linux kernel to complete the GNU system, that boots on a decent
amount of commodity hardware.
I was thinking about 5000 CHF (~4100 USD) I am able to donate myself,
and maybe someone else could join me in this.

Wdyt ?

Claudio

[Alfred M. Szmidt]

       In fact while I am being frank, what is your level of interest
       in the Hurd Richard?  The few times that I have seen you speak
       or read your commentaries, you don't show much of an interest
       yourself.  You speak of Ututo and others when you have legions
       of folks that you could probably drive to the "GNU System".

   When we started development of the Hurd, it was absolutely
   essential.  We needed a free kernel, and there was none.

   Since we have Linux, the Hurd is no longer essential.  Making the
   Hurd work well would be good for two reasons:

   * Because it would be more powerful than Linux.

This isn't true anymore, Linux already does almost everything that the
Hurd does.  Including that Linux has far better, and will always have,
driver support than we can manage.

   * Because it would be a GNU kernel.

   The unfortunate thing is that making the Hurd really secure
   requires Coyotos.  I just asked Marcus Brinkmann how progress is on
   Coyotos.

This isn't true, the Hurd on Mach or L4 can be made equally secure.

[cparker (Bugzilla)-2]

On 7/30/07, Claudio Fontana <claudio.fontana@...> wrote:

> About the last part, maybe some kind of "bounty" could be set up?
> As this is hard work, and maybe not immediately rewarding, maybe
> some tangible monetary reward on a per-feature or per-milestone
> basis could attract new hackers?
>
> I am not familiar with the Hurd and its development status, and do not
> feel able to contribute to its development directly, but I am able to offer a
> moderate bounty for a stable, GPLv3, more or less drop-in replacement
> of the Linux kernel to complete the GNU system, that boots on a decent
> amount of commodity hardware.
> I was thinking about 5000 CHF (~4100 USD) I am able to donate myself,
> and maybe someone else could join me in this.

Although I don't think I'd be able to contribute as much, I'd be more
than willing to contribute toward a Hurd fund. I, too, inquired a
while ago about contributing toward Hurd development, but that didn't
really get anywhere. While I am able to program in C and C++, my
knowledge of kernel development is as great as any random Schmoe on
the street. In lieu of direct participation, if offering funds toward
the completion of the Hurd and the packaging of the GNU system will be
helpful, then I think the FSF/GNU Project should set up a Hurd
development or GNU system packaging fund, similar to what's been done
with the SkyOS "Code Ransom" system: http://www.skyos.org/?q=node/5
(warning: SkyOS is a proprietary operating system).

--
Christopher C. Parker                 Associate Member #795
cparker@...             Free Software Foundation

[Vikram Vincent]

Hello,

On 31/07/07, Christopher Parker <cparker@...> wrote:

On 7/30/07, Claudio Fontana <claudio.fontana@...> wrote:

> About the last part, maybe some kind of "bounty" could be set up?
> As this is hard work, and maybe not immediately rewarding, maybe
> some tangible monetary reward on a per-feature or per-milestone
> basis could attract new hackers?
>
> I am not familiar with the Hurd and its development status, and do not
> feel able to contribute to its development directly, but I am able to offer a
> moderate bounty for a stable, GPLv3, more or less drop-in replacement
> of the Linux kernel to complete the GNU system, that boots on a decent
> amount of commodity hardware.
> I was thinking about 5000 CHF (~4100 USD) I am able to donate myself,
> and maybe someone else could join me in this.

Although I don't think I'd be able to contribute as much, I'd be more
than willing to contribute toward a Hurd fund. I, too, inquired a
while ago about contributing toward Hurd development, but that didn't
really get anywhere. While I am able to program in C and C++, my
knowledge of kernel development is as great as any random Schmoe on
the street. In lieu of direct participation, if offering funds toward
the completion of the Hurd and the packaging of the GNU system will be
helpful, then I think the FSF/GNU Project should set up a Hurd
development or GNU system packaging fund, similar to what's been done
with the SkyOS "Code Ransom" system: http://www.skyos.org/?q=node/5
(warning: SkyOS is a proprietary operating system).

And if the GNU * hackers are from India I'm sure we can try to support  them through our project www.nrcfoss.org.in/
I'm sure that with some reasonable world wide coordination we should be able to move the Hurd development forward faster. Hope the FSF initiates the necessary action.

--
Regards
Vikram Vincent
[phone]
Project Engineer,
NRCFOSS, AU-KBC Research Centre,
http://groups.google.com/group/bangalore_alive/
http://fci.wikia.com/wiki/Campaign_to_promote_free_software_in_karnataka

[arnuld-2]

> On 7/30/07, Richard Stallman <rms@...> wrote:

> When we started development of the Hurd, it was absolutely essential.
> We needed a free kernel, and there was none.

> Since we have Linux, the Hurd is no longer essential.

i don't think so. we can not deal with patents like M$-Novell deal
with GPLv2. corporate have found a way around GPLv2 and  for
contributing to GPLv2 Linux kernel is like dumping the idea of Free
Software. GPLv2 Linux kernel is has no "GPLv2 or later clause" :( and
i do not think that Linus will ever release his kernel under GPLv3. he
will not, as for him GPL does not matter but for me it does. i also
think GPL really matters for GNU community and especially for you
Richard.

[NOTE: everyone is free to call it a biased reply, in fact it is not]


> Making the Hurd
> work well would be good for two reasons:
>
> * Because it would be more powerful than Linux.

true. at the fundamental level it is much powerful an that powerful
kernel is not working as a modern kernel, not yet.

> * Because it would be a GNU kernel.

that is the whole idea i die for  :)


> The unfortunate thing is that making the Hurd really secure
> requires Coyotos.  I just asked Marcus Brinkmann how progress
> is on Coyotos.

after reading the archives of bug-Hurd, help-Hurd and L4-Hurd mailing
lists, i think  it is not true.



--
http://arnuld.blogspot.com/

[arnuld-2]

> On 7/31/07, Vikram Vincent <vincentvikram@...> wrote:

> And if the GNU * hackers are from India I'm sure we can try to support  them
> through our project www.nrcfoss.org.in/
> I'm sure that with some reasonable world wide coordination we should be able
> to move the Hurd development forward faster. Hope the FSF initiates the
> necessary action.

i have read the archives of bug-Hurd, help-Hurd and L4-Hurd and from
that technical discussions and general environment and behaviour of
the folks there i can conclude it is really 100% possible to release
GNU OS and develoep it as fast as GNOME developed and is developing.

in my biased view, we just need to work on organisational problems 1st
and then on specifying to hit a particular target. Hurd folks are
quite helpful and talented and i think we can take this as opportunity
to grab for.


> --
> Regards
> Vikram Vincent
> [phone]
> Project Engineer,
> NRCFOSS, AU-KBC Research Centre,
> http://groups.google.com/group/bangalore_alive/

Bangalore.. hmmm... though Bangalore is more than 3000 km away from
Punjab but i think i can still work with Hurd people there ;-)



--
http://arnuld.blogspot.com/

[Richard Stallman]

It would make sense to set up a Hurd fund if we were confident it
would raise enough money so as to make it possible to spend the money
on the Hurd.  However, I tend to think it would not do so.  Then we
would be stuck with money we could not spend.

Claudio Fontana wrote:

    I was thinking about 5000 CHF (~4100 USD) I am able to donate myself,
    and maybe someone else could join me in this.

That is enough money that it might enable us to start a fund.  However,
we should look for more pledges before actually accepting money.
We would need at least $20,000 to make it useful; that could pay someone
for 6 months.

[Michael Banck-3]

On Mon, Jul 30, 2007 at 08:13:59PM -0400, Christopher Parker wrote:
> Although I don't think I'd be able to contribute as much, I'd be more
> than willing to contribute toward a Hurd fund.

Before jumping the gun, you might want to see how this year's
Hurd-related Google Summer of Code project works out in the end.  


Michael

[Michael Heath]

While I like the ideas of a bounty or pay for developers that has been suggested since this, I think it would be far more appropriate to start with a more basic advertising approach. Up to date, quality documentation and information on the internet is sparse - lots of websites are out of date, and now unmaintained. I know there was talk of starting up a new Hurd wiki, but haven't heard anything since assurances that it would happen a while back.

We need to just start with information that lets people know about the project, answers questions, counters the FUD out there about it, lets people know the conceptual benefits over a kernel like Linux, etc. I'm willing to help with any of this, as much as I can.

Mike Heath

On 7/26/07, Richard Stallman <rms@...> wrote:

    In my opinion, what the Hurd needs right now, and first of all, is
    a maintainer which applies patches and, once that works, some sort of
    advertising to attract new hackers.

I think you are right.

[Michael Heath]

I don't think so. Google's SoC is designed to help finance one student to work on a specific task within a project. With a Hurd-specific fund, you would be able to to choose specifically where to spend the money for it to be the most beneficial to the Hurd. You'd also be able to finance work by developers who may not be elligible for SoC. It's a completely different system.

Mike Heath

On 7/31/07, Michael Banck <mbanck@...> wrote:

On Mon, Jul 30, 2007 at 08:13:59PM -0400, Christopher Parker wrote:
> Although I don't think I'd be able to contribute as much, I'd be more
> than willing to contribute toward a Hurd fund.

Before jumping the gun, you might want to see how this year's
Hurd-related Google Summer of Code project works out in the end.


Michael

[Anders Breindahl]

On 200707311150, Michael Heath wrote:
> I don't think so. Google's SoC is designed to help finance one student to
> work on a specific task within a project. With a Hurd-specific fund, you
> would be able to to choose specifically where to spend the money for it to
> be the most beneficial to the Hurd. You'd also be able to finance work by
> developers who may not be elligible for SoC. It's a completely different
> system.

I agree. If the Hurd wants a kick start towards the end of some of the
goals that are stated -- update glue code, migrate microkernel, make
software package $x compatible with the Hurd (or the reverse), implement
this and that netfs translator, and so on -- paid work would be
justified.

Especially if the FSF or GNU itself will take earmarked donations and
hold them till either used or decidedly returned to donors.

I also agree, that Summer of Code is something different. It's aimed at
smaller projects that fit within the timeframe of one person*summer.
Here we're (rms is) talking about 6 person*months, and projects that
don't need to achieve a stated goal completely.

After all, there isn't so much work that needs to be done, in order for
the Hurd to be a satisfactory standin for Linux. This amount of work
could well do some significant change, in this respect.

Seen from my chair, that is.

Regards, skrewz.