Обсуждение: PGDG RPMS and integer-datetimes support

Can integer datetimes support be added to the PGDG
distributed RPMS for Fedora at the next version
requiring an initdb.  I've attached a diff to the
specfile for 8.0.1.

Thanks, Murthy

--- /usr/src/redhat/SPECS/postgresql-8.0.1-2PGDG.spec
2005-02-22 18:04:50.000000000 -0500
+++
/usr/src/redhat/SPECS/postgresql-8.0.1-2PGDG_BigDT.spec
2005-03-30 13:18:07.528564152 -0500
@@ -67,7 +67,7 @@
 # Pre-release RPM's should not be put up on the
public ftp.postgresql.org server
 # -- only test releases or full releases should be.

-Release: 2PGDG
+Release: 2PGDG_BigDT
 License: BSD
 Group: Applications/Databases
 Source0:
ftp://ftp.postgresql.org/pub/source/v%{version}/postgresql-%{version}.tar.bz2
@@ -365,6 +365,7 @@

 export LIBNAME=%{_lib}
 %configure --disable-rpath \
+    --enable-integer-datetimes \
 %if %beta
     --enable-debug \
     --enable-cassert \





__________________________________
Yahoo! Messenger
Show us what our next emoticon should look like. Join the fun.
http://www.advision.webevents.yahoo.com/emoticontest

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

On Wed, 6 Apr 2005, S Murthy Kambhampaty wrote:

> Can integer datetimes support be added to the PGDG
> distributed RPMS for Fedora at the next version
> requiring an initdb.

Could you please tell us why integer datetimes should be enabled in our
RPMs by default?

We are not sure that many people need it, also it's easy for someone to
add this support using the SRPMs provided.

Regards,
- --
Devrim GUNDUZ
devrim~gunduz.org, devrim~PostgreSQL.org, devrim.gunduz~linux.org.tr
http://www.tdmsoft.com                         http://www.gunduz.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQFCVWEptl86P3SPfQ4RAgYHAJ937NqHHL7VUP7wH7TUnDQ6M1lC4QCg2GwW
dRMyXYCBPG0tfirf53RAG1Y=
=uYoW
-----END PGP SIGNATURE-----

--- Devrim GUNDUZ <devrim@gunduz.org> wrote:
> Could you please tell us why integer datetimes
> should be enabled in our
> RPMs by default?
>
> We are not sure that many people need it, also it's
> easy for someone to
> add this support using the SRPMs provided.
>

"Consistent precision through the range of allowed
values" sceems a feature worth having.

http://archives.postgresql.org/pgsql-hackers/2002-03/msg01038.php

I wonder why you are "not sure that many people need
it".

Regards,
   Murthy



__________________________________
Do you Yahoo!?
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/

S Murthy Kambhampaty <smk_va@yahoo.com> writes:
> "Consistent precision through the range of allowed
> values" sceems a feature worth having.

> I wonder why you are "not sure that many people need
> it".

Because almost nobody has complained about the lack of it.
(I'm talking about actual field experience of there being a
problem, not somebody objecting that it sounds like a
feature worth having.)

It should also be pointed out that we are still finding bugs in
the integer-datetimes code.  This is of course exactly because
it's not the default --- but I feel sure that the average user
who notices a difference at all, if we change the default,
will be much more likely to hit a bug than to benefit.

            regards, tom lane

--- Tom Lane <tgl@sss.pgh.pa.us> wrote:

> Because almost nobody has complained about the lack
> of it.
> (I'm talking about actual field experience of there
> being a
> problem, not somebody objecting that it sounds like
> a
> feature worth having.)
>
> It should also be pointed out that we are still
> finding bugs in
> the integer-datetimes code.  This is of course
> exactly because
> it's not the default --- but I feel sure that the
> average user
> who notices a difference at all, if we change the
> default,
> will be much more likely to hit a bug than to
> benefit.

It certainly did seem like a marginal improvement, but
an improvment nontheless, back when we deployed 7.4 (I
thing the feature was introduced in 7.3).  Now that
we've switched to FC3, it was a minor inconvenience to
have to rebuild the RPM just for this feature, and I
was wondering if there's been enough testing to make
it a default.  Your answer clearly is "no".  Allright.

I wonder if the bugs you're finding are serious enough
to warrant dumping the data and restoring it to a
version without integer-datetimes?

Thanks,
   Murthy




__________________________________
Do you Yahoo!?
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/