The pgreplication project

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

We might be leaning towards migrating things to pgfoundry, but that does
not mean that we can leave gborg behind as a half plundered wreck where
one by accident might still find something useful.

Jan

On 6/4/2004 8:01 AM, Jan Wieck wrote:

Folks,

on the gborg front page is a link to the "pgreplication" project as one
of five "top" projects. This thing has had zero updates since its last
attempts agains PostgreSQL 7.2 and appears literally dead.

Can someone please remove that?

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

Jan,

Perhaps you should contact the project admin or at least
posting to the project mailing list, before assuming the
project is dead. Admittedly there has been several
delays and periods of inactivity, but declaring a
project dead without investigating is quite
inappropriate.

I agree the project should not be in the top 5 on
gborg, but it is not dead! There is a 7.2 port
of the original 6.4 prototype, along with failover
and recovery patches. There is a plan to produce
a full mulit-master replication solution, but I don't
currently have the time to commit to the project. If
there is interest in this type of solution, I'd be
happy to assist in the development.

I've known you for a while, and have a great
deal of respect for you, but I would have
expected some kind of communication with me
before this.

Darren

On 6/8/2004 9:58 AM, darren@up.hrcoxmail.com wrote:

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

Jan,

Perhaps you should contact the project admin or at least
posting to the project mailing list, before assuming the
project is dead. Admittedly there has been several
delays and periods of inactivity, but declaring a
project dead without investigating is quite
inappropriate.

I apologize and admit that it should have been posted to the project
mailing list.

I agree the project should not be in the top 5 on
gborg, but it is not dead! There is a 7.2 port
of the original 6.4 prototype, along with failover
and recovery patches. There is a plan to produce
a full mulit-master replication solution, but I don't
currently have the time to commit to the project. If
there is interest in this type of solution, I'd be
happy to assist in the development.

Do those plans include fixing the problems arising from the complete
reversal of lock priorities with aborting previous lock holders? I have
stated before and do it again that this part of the Postgres-R logic is
assuming a more or less lock conflict free application design, which is
in practice seldom (if ever) found.

I've known you for a while, and have a great
deal of respect for you, but I would have
expected some kind of communication with me
before this.

I fully return the same respect to you. I was not aware that you are
still feeling "in charge" of this project.

Sincerely,
Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

Jan Wieck wrote:

On 6/8/2004 9:58 AM, darren@up.hrcoxmail.com wrote:

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

Jan,

Perhaps you should contact the project admin or at least
posting to the project mailing list, before assuming the
project is dead. Admittedly there has been several
delays and periods of inactivity, but declaring a
project dead without investigating is quite
inappropriate.

I apologize and admit that it should have been posted to the project
mailing list.

I assumed the top-5 projects are dynamic based on the amount of
activity, not hardcoded as they seem to be.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

On 6/8/2004 11:04 AM, Bruce Momjian wrote:

Jan Wieck wrote:

On 6/8/2004 9:58 AM, darren@up.hrcoxmail.com wrote:

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

Jan,

Perhaps you should contact the project admin or at least
posting to the project mailing list, before assuming the
project is dead. Admittedly there has been several
delays and periods of inactivity, but declaring a
project dead without investigating is quite
inappropriate.

I apologize and admit that it should have been posted to the project
mailing list.

I assumed the top-5 projects are dynamic based on the amount of
activity, not hardcoded as they seem to be.

Looking at "orapgsqlviews", which had the last CVS commit 3 years ago
and is still clinging to its top-5 position, I don't think so. Chris?

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

Well I am working to address the problem, sorry if I am not doing it
fast enough. I spoke with the gborg maintainer and found out that the
top 5 is based on page views for a project rather than cvs activity.
While I've no doubt you've been more active on slony development, I am
not so sure that pgreplication doesn't get more page views. So, I have
(just) sent Chris an email asking to reset the stats so we will see
which project floats to the top, but I think longer term we need a
"replication guide" listing the different solutions available like we do
with windows and gui tools up on techdocs. I plan to do that as well,
just need to find time to work it into the schedule.

Robert Treat

On Tue, 2004-06-08 at 08:46, Jan Wieck wrote:

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

We might be leaning towards migrating things to pgfoundry, but that does
not mean that we can leave gborg behind as a half plundered wreck where
one by accident might still find something useful.

Jan

On 6/4/2004 8:01 AM, Jan Wieck wrote:

Folks,

on the gborg front page is a link to the "pgreplication" project as one
of five "top" projects. This thing has had zero updates since its last
attempts agains PostgreSQL 7.2 and appears literally dead.

Can someone please remove that?

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faqs/FAQ.html

--
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL

On 6/8/2004 11:49 AM, Robert Treat wrote:

Well I am working to address the problem, sorry if I am not doing it
fast enough. I spoke with the gborg maintainer and found out that the
top 5 is based on page views for a project rather than cvs activity.

Aha ... and let me guess, the algorithm does NOT look at the
Http-Referrer to ignore the hits caused by listing it at the top, right?

While I've no doubt you've been more active on slony development, I am
not so sure that pgreplication doesn't get more page views. So, I have
(just) sent Chris an email asking to reset the stats so we will see
which project floats to the top, but I think longer term we need a
"replication guide" listing the different solutions available like we do
with windows and gui tools up on techdocs. I plan to do that as well,
just need to find time to work it into the schedule.

Thanks,

Jan

Robert Treat

On Tue, 2004-06-08 at 08:46, Jan Wieck wrote:

I guess I have to address a slightly broader audience, or do I conclude
from the total lack of interest in this matter that nobody cares if we
promote something here as one of our "top five" flagship projects, that
is inactive for years and still a prototype implementation against
PostgreSQL 6.4 with no ideas how to incorporate the semantic changes of
MVCC into the concept.

We might be leaning towards migrating things to pgfoundry, but that does
not mean that we can leave gborg behind as a half plundered wreck where
one by accident might still find something useful.

Jan

On 6/4/2004 8:01 AM, Jan Wieck wrote:

Folks,

on the gborg front page is a link to the "pgreplication" project as one
of five "top" projects. This thing has had zero updates since its last
attempts agains PostgreSQL 7.2 and appears literally dead.

Can someone please remove that?

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faqs/FAQ.html

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #

Jan,

The Top projects list is a dynamic list based on a cumulative page
views. Occassionally the ranking is scalled or zero'd to give newer
projects a chance to compete for the top rankings with older more
established projects. This last occurred a few months ago.

Hope this answers your question.

Chris Ryan

--- Jan Wieck <JanWieck@Yahoo.com> wrote:

On 6/8/2004 11:04 AM, Bruce Momjian wrote:

Jan Wieck wrote:

On 6/8/2004 9:58 AM, darren@up.hrcoxmail.com wrote:

I guess I have to address a slightly broader audience, or do I

conclude

from the total lack of interest in this matter that nobody

cares if we

promote something here as one of our "top five" flagship

projects, that

is inactive for years and still a prototype implementation

against

PostgreSQL 6.4 with no ideas how to incorporate the semantic

changes of

MVCC into the concept.

Jan,

Perhaps you should contact the project admin or at least
posting to the project mailing list, before assuming the
project is dead. Admittedly there has been several
delays and periods of inactivity, but declaring a
project dead without investigating is quite
inappropriate.

I apologize and admit that it should have been posted to the

project

mailing list.

I assumed the top-5 projects are dynamic based on the amount of
activity, not hardcoded as they seem to be.

Looking at "orapgsqlviews", which had the last CVS commit 3 years ago

and is still clinging to its top-5 position, I don't think so. Chris?

Jan

--

#======================================================================#

# It's easier to get forgiveness for being wrong than for being
right. #
# Let's break this rule - forgive me.
#
#==================================================
JanWieck@Yahoo.com #

__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/

Bruce,

I assumed the top-5 projects are dynamic based on the amount of
activity, not hardcoded as they seem to be.

Well, this is the issue ... they are not hardcoded, but they are based
entirely on the number of page views, and not other forms of activity. this
includes page views referred from the "top 5 list" itself ... which means
that once you're in the top 5, it tends to be self-perpetuating. Chris has
been attacking this manually by re-setting the counter every month.

I'm not sure that pgFoundry solves that problem. The issue is that
"activity" *should* be the count of:
cvs check-ins and check-outs
mailing list posts
forum posts
news items
downloads
bug reports

The problem is that pgFoundry does not have any meaningful way to measure most
of the above; it's somewhat disconnected from the mailman archives, and only
acts as a gateway to CVS. Downloads are frequently mirrored. I think Chris
is in the same bind with GBorg.

I will dig into the GForge code for "top projects" to see how it does work,
though, when I have a chance.

--
Josh Berkus
Aglio Database Solutions
San Francisco

Cumulative page views number is a wrong parameter for selecting the
top projects list IMHO. May be the number of downloads?

Regards,
Mikhail Terekhov

Chris Ryan wrote: