Re: UDD data - MySQL - Was: Re: Archiving UDD data : FLOSSMole ?

2009-06-17 Thread Stefano Zacchiroli
On Tue, Jun 16, 2009 at 11:22:10PM +0200, Olivier Berger wrote:
 I think it depends the kind of analysis people might want to do for
 their research in mining the Debian metadata that would be archived in
 FLOSSMole, maybe these package versions wouldn't be needed, so strings
 without the full arithmetics of versions would be enough ?
 
 In any case, the first idea is to archive data so that it's there for
 history, I suppose, then the users will complain eventually, and we
 shall see ;)

Exactly. I see no problem in archiving it to MySQL, injecting versions
as string. That would just mean that you can't run the exactly same
query you run now, but you can still re-inject historical versions
elsewhere and/or use some MySQL extensions to obtain the same effect.

Cheers.

-- 
Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc @ Univ. Paris 7
z...@{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
sempre uno zaino ...| ..: | Je dis tu à tous ceux que j'aime


signature.asc
Description: Digital signature


Re: UDD data - MySQL - Was: Re: Archiving UDD data : FLOSSMole ?

2009-06-16 Thread Stefano Zacchiroli
On Wed, Jun 10, 2009 at 05:59:21PM +0200, Olivier Berger wrote:
 Actually, something obvious seems to render things a bit difficult at
 first sight : UDD is in PostGres, and FLOSSMole uses MySQL.
 
 We're discussing some options on the ossmole-discuss list to overcome
 this difficulty.
 
 My advice is to import in PG on a Debian system with the required
 dependencies, create views to be able to convert non standard types to
 standard ones, then dump again to SQL, and import to MySQL... any
 comments ?

UDD has some parts that are Postgres specific, in particular it uses
the Postgres extension mechanism to internalize Debian version
comparison so that it can be exploited in queries. In the beginning,
the internalization was done only as an extension function, it might
be that now there is even a custom data type where you can use  and
such, but I'm not sure about that (redirect to Lucas).

You just need to pay attention (and maybe do some tests) that,
migrating away from Postgres, you might loose the ability to reuse the
same queries which you can currently use with udd.d.o.

Cheers.

-- 
Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc @ Univ. Paris 7
z...@{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
sempre uno zaino ...| ..: | Je dis tu à tous ceux que j'aime


signature.asc
Description: Digital signature


Re: UDD data - MySQL - Was: Re: Archiving UDD data : FLOSSMole ?

2009-06-16 Thread Andreas Tille
On Tue, Jun 16, 2009 at 09:49:27AM +0200, Stefano Zacchiroli wrote:
 UDD has some parts that are Postgres specific, in particular it uses
 the Postgres extension mechanism to internalize Debian version
 comparison so that it can be exploited in queries. In the beginning,
 the internalization was done only as an extension function, it might
 be that now there is even a custom data type where you can use  and
 such, but I'm not sure about that (redirect to Lucas).

I can confirm that there is a new data type debversion:

udd= \d packages
   Table public.packages
   Column|Type| Modifiers 
-++---
 package | text   | not null
 version | debversion | not null
 architecture| text   | not null
 maintainer  | text   | 
 description | text   | 
 long_description| text   | 
 source  | text   | 
 source_version  | debversion | 
...

which is probably hard to reimplement in MySQL.  There was a thread
here on this list about this data type and its implementation.

 You just need to pay attention (and maybe do some tests) that,
 migrating away from Postgres, you might loose the ability to reuse the
 same queries which you can currently use with udd.d.o.

IMHO it is not a good idea to try to port UDD to a different
database engine.

Kind regards

Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: UDD data - MySQL - Was: Re: Archiving UDD data : FLOSSMole ?

2009-06-16 Thread Olivier Berger
Le mardi 16 juin 2009 à 12:50 +0200, Andreas Tille a écrit :
 On Tue, Jun 16, 2009 at 09:49:27AM +0200, Stefano Zacchiroli wrote:
  UDD has some parts that are Postgres specific, in particular it uses
  the Postgres extension mechanism to internalize Debian version
  comparison so that it can be exploited in queries. In the beginning,
  the internalization was done only as an extension function, it might
  be that now there is even a custom data type where you can use  and
  such, but I'm not sure about that (redirect to Lucas).
 
 I can confirm that there is a new data type debversion:
 

SNIP

 which is probably hard to reimplement in MySQL.  There was a thread
 here on this list about this data type and its implementation.
 

Thanks for these comments.

That's what I had also indentified as PostGres dependant when we
discussed the subject on the ossmole-discuss list.

  You just need to pay attention (and maybe do some tests) that,
  migrating away from Postgres, you might loose the ability to reuse the
  same queries which you can currently use with udd.d.o.
 
 IMHO it is not a good idea to try to port UDD to a different
 database engine.
 

I think it depends the kind of analysis people might want to do for
their research in mining the Debian metadata that would be archived in
FLOSSMole, maybe these package versions wouldn't be needed, so strings
without the full arithmetics of versions would be enough ?

In any case, the first idea is to archive data so that it's there for
history, I suppose, then the users will complain eventually, and we
shall see ;)

Thanks again.

Best regards,
-- 
Olivier BERGER olivier.ber...@it-sudparis.eu
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 1024D/6B829EEC
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)


--
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



UDD data - MySQL - Was: Re: Archiving UDD data : FLOSSMole ?

2009-06-10 Thread Olivier Berger
Hi.

Le mardi 09 juin 2009 à 16:55 +0200, Olivier Berger a écrit :

 It's very likely then, that UDD will start being collected by the
 FLOSSMole team in the future.
 

Actually, something obvious seems to render things a bit difficult at
first sight : UDD is in PostGres, and FLOSSMole uses MySQL.

We're discussing some options on the ossmole-discuss list to overcome
this difficulty.

My advice is to import in PG on a Debian system with the required
dependencies, create views to be able to convert non standard types to
standard ones, then dump again to SQL, and import to MySQL... any
comments ?

Thanks in advance.

Best regards,
-- 
Olivier BERGER olivier.ber...@it-sudparis.eu
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 1024D/6B829EEC
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)


--
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: Archiving UDD data : FLOSSMole ?

2009-06-09 Thread Olivier Berger
Le lundi 08 juin 2009 à 14:35 +0200, Olivier Berger a écrit :
 Le dimanche 07 juin 2009 à 08:43 +0200, Stefano Zacchiroli a écrit :
  
  Even though it will not solve that problem, I would very welcome
  injecting periodically data from UDD to the FLOSSMole guys. The
  simplest way, currently, seems to be to handing hover the snapshot we
  periodically release on the web. Have you already asked them if that
  kind of data would be suitable for them?
  
 
 More or less, orally, and at least Megan Squire was quite receptive (I
 actually wrote my mail during her talk's Q  As). I will try and post to
 their list and report here.
 
 I'm not exactly sure of all the constraints involved with archiving some
 data at their place, but will try and act as a facilitator, and report
 when I have more details, letting you probably discuss further
 details ;)
 

Following-up on this, you'll find the thread around
http://sourceforge.net/mailarchive/forum.php?thread_name=94d3bd120906090719p3b1000f2u9d1e160cd1fe064a%40mail.gmail.comforum_name=ossmole-discuss
 
where the idea seems accepted.

It's very likely then, that UDD will start being collected by the
FLOSSMole team in the future.

I'm staying subscribed both to debian-qa and flossmole-discuss lists to
try and proxy as much information as would be possible, but feel free to
take direct contacts if you see fit ;)

Best regards,
-- 
Olivier BERGER olivier.ber...@it-sudparis.eu
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 1024D/6B829EEC
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)


--
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: Archiving UDD data : FLOSSMole ?

2009-06-08 Thread Olivier Berger
Le dimanche 07 juin 2009 à 08:43 +0200, Stefano Zacchiroli a écrit :
 
 Even though it will not solve that problem, I would very welcome
 injecting periodically data from UDD to the FLOSSMole guys. The
 simplest way, currently, seems to be to handing hover the snapshot we
 periodically release on the web. Have you already asked them if that
 kind of data would be suitable for them?
 

More or less, orally, and at least Megan Squire was quite receptive (I
actually wrote my mail during her talk's Q  As). I will try and post to
their list and report here.

I'm not exactly sure of all the constraints involved with archiving some
data at their place, but will try and act as a facilitator, and report
when I have more details, letting you probably discuss further
details ;)

Btw, I'm not sure if/how we can plan some UDD-related discussions
@debconf... not having participated to debconf earlier... I suppose
we'll manage to find a place with cervesa to do so in last extremity ;)

Best regards,
-- 
Olivier BERGER olivier.ber...@it-sudparis.eu
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 1024D/6B829EEC
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)


--
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: Archiving UDD data : FLOSSMole ?

2009-06-08 Thread Lucas Nussbaum
On 08/06/09 at 14:35 +0200, Olivier Berger wrote:
 Btw, I'm not sure if/how we can plan some UDD-related discussions
 @debconf... not having participated to debconf earlier... I suppose
 we'll manage to find a place with cervesa to do so in last extremity ;)

Hi,

I proposed a UDD talk at debconf, which was accepted. So this could be
an opportunity to find people interested in UDD, and then have offline
discussions.

However, I'm not 100% sure yet that I will come to Debconf, as I might
need the time to move to another city if I choose the position I'm being
offered there.

Also, sorry again for not being able to respond quickly to all
UDD-related emails and questions: I'm currently kept totally busy by
real life stuff. I realize it is very bad timing, but there's not much I
can do :(
-- 
| Lucas Nussbaum
| lu...@lucas-nussbaum.net   http://www.lucas-nussbaum.net/ |
| jabber: lu...@nussbaum.fr GPG: 1024D/023B3F4F |


-- 
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: Archiving UDD data : FLOSSMole ?

2009-06-08 Thread Andreas Tille
On Mon, Jun 08, 2009 at 10:28:37PM +0200, Lucas Nussbaum wrote:
 On 08/06/09 at 14:35 +0200, Olivier Berger wrote:
  Btw, I'm not sure if/how we can plan some UDD-related discussions
  @debconf... not having participated to debconf earlier... I suppose
  we'll manage to find a place with cervesa to do so in last extremity ;)
 
 I proposed a UDD talk at debconf, which was accepted. So this could be
 an opportunity to find people interested in UDD, and then have offline
 discussions.

I'm definitely interested and my talk about the usage of UDD in
Debian Pure Blends will be also UDD related.  I'd propose an ad hoc
BOF for those who are interested.
 
 However, I'm not 100% sure yet that I will come to Debconf, as I might
 need the time to move to another city if I choose the position I'm being
 offered there.

Would be a shame to miss you.

Kind regards

 Andreas. 

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: Archiving UDD data : FLOSSMole ?

2009-06-07 Thread Stefano Zacchiroli
On Sat, Jun 06, 2009 at 12:55:39PM +0200, Olivier Berger wrote:
 I think maybe it would be interesting to offer some data from UDD to
 the FLOSSMole research archive at :
 http://ossmole.sourceforge.net/ (and
 http://code.google.com/p/flossmole/ ). By offering, I mean there may be
 some need of synchronisation with the researchers for technical matters
 (or others : anonymization, etc.)...

Very interesting! thanks for the heads up, as I wasn't aware of that
project in spite of having need in the past such kind of historical
data.

Actually, during the design of UDD, I've pushed a bit to have *some*
historical data, in particular I was interested in having all the
upload history. Back then however, it turned out that it was too
heavy to have that naively in the database itself.

Even though it will not solve that problem, I would very welcome
injecting periodically data from UDD to the FLOSSMole guys. The
simplest way, currently, seems to be to handing hover the snapshot we
periodically release on the web. Have you already asked them if that
kind of data would be suitable for them?

 Any comments ?
 I hope to meet some of you at Debconf to talk to you about that.

I'll be there, and I'll very happily join a discussion about this.

Cheers.

-- 
Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc @ Univ. Paris 7
z...@{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
sempre uno zaino ...| ..: | Je dis tu à tous ceux que j'aime


signature.asc
Description: Digital signature