Stefano Zacchiroli | 2 May 10:40 2010
Picon

Re: Estadistical proyect.

On Wed, Apr 21, 2010 at 10:58:15AM +0200, angel wrote:
> My name is Angel Bravo, I'm a student at the University of Granada,
> (Spain) in his final year of a masters degree in Statistics. In order to
> finish my degree I am supposed to do a major work in statistics. In case
> Debian is interested in having someone work on a statistical project of
> any nature (the project would be supervised by professors of the
> Statistics and Operations Research department of the University of
> Granada and and Debian would not be charged with any costs whatsoever),
> please feel free to contact me. Yours sincerely, Angel Bravo. 

Dear Angel,
  we do are interested in principle, but I confess your offer is a bit
too vague to understand what you're actually interested in doing. Some
people has shown interest in your proposal (I'm Cc-ing a couple of
them), and I'm personally interested in whatever statistical results
which might raise our awareness on any specific problem we might have
(on manpower, on software bugs, on timing of our procedures, etc.).

For instance, Gaudenz (Cc-ed) has in the past analyzed the community of
Debian contributors, classified them in different categories, and
established a lot of interesting fact on how/when they work, how they
evolve over time, etc. From that, we have been able to derive some very
useful information. That is just an example of work in which statistics
was involved which has been useful to us.

However, we don't have any specific "assignment" to offer right now,
maybe you want to think a bit more about a specific topic and re-iterate
your request? We'll be more than happy to help, once the proposal is a
bit more clear.

(Continue reading)

Martin Zobel-Helas | 2 May 16:45 2010
Picon

Proposed changes to the Debian Machine Usage Policy (DMUP)

Hi,

the Debian Project Leader recently asked DSA in his delegation[1] to fix
a flaw in the current version of the DMUP, the fact that expulsions are
DAM's domain, and not our's.

I took the opportunity to also fix some other minor stuff in the current
version of the document. Please note that this is a first draft, not the
final version, which needs to be announced as per delegation to
debian-devel-announce.

Summary of changes: The Debian Systems Administration Team will do
whatever is necessary to keep all machines and services running.

Or to say it with someone else's words from IRC:
"DSA has to look after this crap, so please don't be a dick."

Find attached my proposed changes as commit log, as diff to the current
version and as full text of the new version.

I started mainaining the new version in git on
http://db.debian.org/git/DMUP.git, copied over frequently to
http://git.debian.org/?p=mirror/DMUP.git;a=summary (gitweb frontend).

Cheers,
Martin

[1] http://lists.debian.org/debian-devel-announce/2010/04/msg00016.html
--

-- 
 Martin Zobel-Helas <zobel <at> debian.org>  | Debian System Administrator
(Continue reading)

angel | 2 May 17:27 2010
Picon

Re: Estadistical proyect.

El dom, 02-05-2010 a las 10:40 +0200, Stefano Zacchiroli escribió:
> On Wed, Apr 21, 2010 at 10:58:15AM +0200, angel wrote:
> > My name is Angel Bravo, I'm a student at the University of Granada,
> > (Spain) in his final year of a masters degree in Statistics. In order to
> > finish my degree I am supposed to do a major work in statistics. In case
> > Debian is interested in having someone work on a statistical project of
> > any nature (the project would be supervised by professors of the
> > Statistics and Operations Research department of the University of
> > Granada and and Debian would not be charged with any costs whatsoever),
> > please feel free to contact me. Yours sincerely, Angel Bravo. 
> 
> Dear Angel,
>   we do are interested in principle, but I confess your offer is a bit
> too vague to understand what you're actually interested in doing. Some
> people has shown interest in your proposal (I'm Cc-ing a couple of
> them), and I'm personally interested in whatever statistical results
> which might raise our awareness on any specific problem we might have
> (on manpower, on software bugs, on timing of our procedures, etc.).
> 
> For instance, Gaudenz (Cc-ed) has in the past analyzed the community of
> Debian contributors, classified them in different categories, and
> established a lot of interesting fact on how/when they work, how they
> evolve over time, etc. From that, we have been able to derive some very
> useful information. That is just an example of work in which statistics
> was involved which has been useful to us.
> 
> However, we don't have any specific "assignment" to offer right now,
> maybe you want to think a bit more about a specific topic and re-iterate
> your request? We'll be more than happy to help, once the proposal is a
> bit more clear.
(Continue reading)

Stefano Zacchiroli | 2 May 18:08 2010
Picon

Re: Proposed changes to the Debian Machine Usage Policy (DMUP)

On Sun, May 02, 2010 at 04:45:42PM +0200, Martin Zobel-Helas wrote:
> Find attached my proposed changes as commit log, as diff to the current
> version and as full text of the new version.

I've just a main comment on your proposed changes (and which relates to
the limit stated in the delegation): I think DSA should be free to
terminate/lock accounts as needed, what they should rather only propose
to DAM is the expulsion to the project. I.e., I see a difference between
"having a (working) account" and "being a project member". So, in
paragraphs like:

> +              3. If the offender fails to contact DSA within 14 days, DSA
> +                 will propose this account to DAM to be terminated and the
> +                 offender expelled from the Debian project. If the offender
> +                 has announced they will be on vacation in this time frame
> +                 this period will be extended with the announced duration of
> +                 the vacation.

and

> +                 DSA will propose this account to DAM to be terminated
> +                 immediately.

I will clarify that what will be be proposed to DAM is expulsion, not
account locking/termination (which is usual sysadm business and I don't
see why DSA should not have that power).

In case you like this change, please find attached a patch against the
current v1.1.2 branch.

(Continue reading)

Stefano Zacchiroli | 2 May 18:18 2010
Picon

Re: Estadistical proyect.

On Sun, May 02, 2010 at 05:27:21PM +0200, angel wrote:
> It's not a bit, it's too much vague :) but for two basic reasons:
> 	
> .- The project can be as big (or hard) as we need (always within range
> of my master degree, of course, only must to study it), so you are
> completely free for to propose me any idea or project (like manpower, on
> software bugs, on timing of our procedures, etc...). Is like a blank
> canvas...but you must to say me what exactly want I'll do with the data
> (for example, something like obtain a certain probability/stadistic) I
> need to known the data and what do you want I do with these data to
> start and I'm sure you know this much better than me.

Then, I fear, we won't have a deal :-) We generally don't work this way:
while we welcome any kind of input and help, the proposer usually needs
to show some pro-activity and autonomy.

Additionally, it seems that currently we don't have any specific idea
about the "precise assignment" you seem to be looking for. So, unless
someone steps in this thread with that (and maybe with willingness to
guide you), you'll have to come up with a precise proposal yourself and
some willingness to dig into our data which, by the way, are publicly
available and generally quite accessible (e.g. in UDD:
http://udd.debian.org).

Cheers.

--

-- 
Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc  <at>  Univ. Paris 7
zack <at> {upsilon.cc,pps.jussieu.fr,debian.org} -<>- http://upsilon.cc/zack/
Dietro un grande uomo c'è ..|  .  |. Et ne m'en veux pas si je te tutoie
(Continue reading)

angel | 2 May 19:29 2010
Picon

Re: Estadistical proyect.

I see...I don't understand exactly what are the meaning of
"pro-activity" or "autonomy" for you, maybe cause never do something
like a software contribution, and feel a bit lost, sorry :) but you are
right when said I need someone who oriented and tell me what "debian
need". I'll try to do "something useful" with your link to the debian
database (thanks a lot), but if someone got a better idea or think that
there is anything better I could do, contact with me. Thanks Stefano.

El dom, 02-05-2010 a las 18:18 +0200, Stefano Zacchiroli escribió:
> On Sun, May 02, 2010 at 05:27:21PM +0200, angel wrote:
> > It's not a bit, it's too much vague :) but for two basic reasons:
> > 	
> > .- The project can be as big (or hard) as we need (always within range
> > of my master degree, of course, only must to study it), so you are
> > completely free for to propose me any idea or project (like manpower, on
> > software bugs, on timing of our procedures, etc...). Is like a blank
> > canvas...but you must to say me what exactly want I'll do with the data
> > (for example, something like obtain a certain probability/stadistic) I
> > need to known the data and what do you want I do with these data to
> > start and I'm sure you know this much better than me.
> 
> Then, I fear, we won't have a deal :-) We generally don't work this way:
> while we welcome any kind of input and help, the proposer usually needs
> to show some pro-activity and autonomy.
> 
> Additionally, it seems that currently we don't have any specific idea
> about the "precise assignment" you seem to be looking for. So, unless
> someone steps in this thread with that (and maybe with willingness to
> guide you), you'll have to come up with a precise proposal yourself and
> some willingness to dig into our data which, by the way, are publicly
(Continue reading)

Steffen Möller | 2 May 23:50 2010
Picon
Picon

Re: Estadistical proyect.

Hello,

for the truly interesting statistics with Debian itself we don't
have the data, and we don't want to have it since we don't want
our users to be possibly identified and characterised in some way.

But if you think a bit upstream of Debian, we can definitely need
every minute of yours. With my education-hat on, I suggest that
you have a look at OpenOffice (or kspread or gnumeric to be
politically correct) and then check out its statistical functions.
You will find them basically not to exist and to be absolutely
impossible (well, almost) to set the parameters.

Your master work could identify ways to bring regular spreadsheets
closer to what various statsoft programs are already providing for us.
And then you should implement at least some skeleton of it. You have
some six months, right? If it was easy, then Redmond would have
done something to its hilariously bad corner of Excel.

I'd fancy an integration of R. So maybe you could somehow map the
two suites.

Just an idea. Debian brings software and skills to work with it
to the end users and developers alike. Many of us are also involved
in some upstream project. You could help with manual/tutorial
writing any time, also at the core of Debian, but not for your master thesis.

Many greetings

Steffen (two weeks almost offline)
(Continue reading)

angel | 3 May 00:26 2010
Picon

Re: Estadistical proyect.

Hello

Sorry, but it seems you are wrong :-) (or maybe I couldn't explain as
well as I thought). I don't want to do any thesis or use the core of
debian for it, just want to take the opportunity of make a real, free
and high level stadistical proyect (and supervise by teachers) for a
small contribution for Debian...this is not my "six months master work"
and obviously we are talking about something wich it's in a different
level where Openoffice, kspread or gnumeric are useless. Maybe to
propose this here was a big mistake, not happen again.

Many greetings

El dom, 02-05-2010 a las 23:50 +0200, Steffen Möller escribió:
> Hello,
> 
> for the truly interesting statistics with Debian itself we don't
> have the data, and we don't want to have it since we don't want
> our users to be possibly identified and characterised in some way.
> 
> But if you think a bit upstream of Debian, we can definitely need
> every minute of yours. With my education-hat on, I suggest that
> you have a look at OpenOffice (or kspread or gnumeric to be
> politically correct) and then check out its statistical functions.
> You will find them basically not to exist and to be absolutely
> impossible (well, almost) to set the parameters.
> 
> Your master work could identify ways to bring regular spreadsheets
> closer to what various statsoft programs are already providing for us.
> And then you should implement at least some skeleton of it. You have
(Continue reading)

Alexander Reichle-Schmehl | 3 May 09:55 2010

Re: Proposed changes to the Debian Machine Usage Policy (DMUP)

Hi!

On 02.05.2010 16:45, Martin Zobel-Helas wrote:

> the Debian Project Leader recently asked DSA in his delegation[1] to fix
> a flaw in the current version of the DMUP, the fact that expulsions are
> DAM's domain, and not our's.

Could you please explain, why this is a flaw?

Best regards,
   Alexander

Stefano Zacchiroli | 3 May 10:45 2010
Picon

Re: Proposed changes to the Debian Machine Usage Policy (DMUP)

On Mon, May 03, 2010 at 09:55:31AM +0200, Alexander Reichle-Schmehl wrote:
> Could you please explain, why this is a flaw?

I think that I should do that, but first let me quote the "Comments"
section of the delegation text, as it was precisely my attempt at
explaining that:

> The ability to change DMUP is meant to allow the document to evolve
> with the addition of new services, technologies, security needs and
> procedures, etc.
>
> The first proposed limit to what can be changed by DSA is meant to fix
> a "flaw" in the current text. Decisions over Debian membership are
> already a responsibility of DAM and, especially considering the need
> of changing the DMUP, it is better not to mix that responsibility with
> DSA (as a paranoid mind can imagine DSA changing DMUP *precisely* to
> have a specific developer expelled ...).  Note that this limit means
> that the current DMUP text is outside the rules and should hence be
> fixed ASAP.

The full context (also discussing the ability the project have to
counter the risk of DSA-going-mad :)) is at [1].

If that is not deemed to be clear enough or satisfactory, I'm happy to
clarify any additional doubt that people might have.

Cheers.

[1] http://lists.debian.org/debian-devel-announce/2010/04/msg00016.html

(Continue reading)


Gmane