Thread: All three packages ...
I never got any feedback on v7.3.7's packaging, so I'm guessing that it looked okay to everyone, even though it was substantially smaller? The only two things that I can think of that might have failed appear to be in there: %tar tvzpf postgresql-7.3.7.tar.gz | grep man.tar -rw-r--r-- pgsql/wheel 146009 Aug 17 16:18 2004 postgresql-7.3.7/doc/man.tar.gz %tar tvzpf postgresql-7.3.7.tar.gz | grep postgres.tar -rw-r--r-- pgsql/wheel 956469 Aug 17 16:18 2004 postgresql-7.3.7/doc/postgres.tar.gz I just got 7.2.5 done now ... I pulled the man.tar.gz file from 7.2.4 and put a copy of it into ~ftp/pub/dev/doc/man-7.2.tar.gz (in case its needed in the future) ... Devrim/David ... I'm doing a force sync of the ftp site from developer -> ftp right now, so all the bundles should be there for you ... the SRPMs are done for 7.4.5, are you going to do similiar with the 7.2.5 and 7.3.7 distros? Baring any comments about problem with the packages from anyone, I'll do a general announce of these releases later on this evening ... ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> I never got any feedback on v7.3.7's packaging, so I'm guessing that it
> looked okay to everyone, even though it was substantially smaller?
Devrim pointed out yesterday that it seemed to be missing
contrib/earthdistance; that's not enough to explain the size diff
though.
regards, tom lane
Nope, and I even checked jdbc to make sure nothing go screwed up with the moves post-7.3, and its there also ... and 7.2 *appears* to be okay size wise ... On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> I never got any feedback on v7.3.7's packaging, so I'm guessing that it >> looked okay to everyone, even though it was substantially smaller? > > Devrim pointed out yesterday that it seemed to be missing > contrib/earthdistance; that's not enough to explain the size diff > though. > > regards, tom lane > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> I just got 7.2.5 done now ... I pulled the man.tar.gz file from 7.2.4 and
> put a copy of it into ~ftp/pub/dev/doc/man-7.2.tar.gz (in case its needed
> in the future) ...
The 7.2.5 tarball is considerably bigger than 7.2.4; this seems to be
because the built .html files are laying about in doc/src/sgml, instead
of having been deleted again ...
regards, tom lane
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> I never got any feedback on v7.3.7's packaging, so I'm guessing that it
> looked okay to everyone, even though it was substantially smaller?
I diff'd the 7.3.6 and 7.3.7 tarballs, and got these unexpected
differences:
Only in postgresql-7.3.6/contrib/earthdistance: Makefile
Only in postgresql-7.3.6/contrib/earthdistance: README.earthdistance
Only in postgresql-7.3.6/contrib/earthdistance: earthdistance.c
Only in postgresql-7.3.6/contrib/earthdistance: earthdistance.sql.in
Only in postgresql-7.3.6/contrib/earthdistance: expected
Only in postgresql-7.3.6/contrib/earthdistance: sql
Only in postgresql-7.3.7: postgresql-base-7.3.7.tar.gz.md5
Only in postgresql-7.3.7: postgresql-docs-7.3.7.tar.gz.md5
Only in postgresql-7.3.7: postgresql-opt-7.3.7.tar.gz.md5
Only in postgresql-7.3.7: postgresql-test-7.3.7.tar.gz.md5
The lack of earthdistance needs to be fixed (checked out wrong module
no doubt?) The extra md5 files aren't that important but maybe could
be cleaned out when you rebuild.
Otherwise it looks good, including the docs.
regards, tom lane
One other little problem with the 7.2.5 package: I see you rebuilt the
flex and bison output files with our current versions of those tools.
I think this is probably okay, but it changes the expected output of
some of the regression tests, eg
*** ./expected/strings.out Fri Jun 1 13:49:17 2001
--- ./results/strings.out Thu Aug 19 12:08:22 2004
***************
*** 17,23 **** ' - next line' /* this comment is not allowed here */ ' - third line' AS "Illegal comment within
continuation";
! ERROR: parser: parse error at or near "'" -- -- test conversions between various string types --
--- 17,23 ---- ' - next line' /* this comment is not allowed here */ ' - third line' AS "Illegal comment within
continuation";
! ERROR: parser: syntax error at or near "'" -- -- test conversions between various string types --
A really paranoid person might complain that we have not adequately
tested that the 7.2 code base will work with the updated tools.
I don't think this is a big deal (I did run through the regression
tests to verify this...) but we will need to fix the regression
expected outputs if we want to do it this way.
If you want to proceed, let me know and I'll commit the updated
regression outputs into the 7.2 branch, and then we can re-wrap.
regards, tom lane
The good news is that the 7.4.5 package looks right ;-)
regards, tom lane
I've just done a re-export of teh CVS for 7.3.7 ... this time, earthdistance appears to be in there properly ... not sure what happened, but am building a new package for it, and will check sizes again when finished ... On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> I never got any feedback on v7.3.7's packaging, so I'm guessing that it >> looked okay to everyone, even though it was substantially smaller? > > Devrim pointed out yesterday that it seemed to be missing > contrib/earthdistance; that's not enough to explain the size diff > though. > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 8: explain analyze is your friend > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
On Thu, 19 Aug 2004, Tom Lane wrote: > One other little problem with the 7.2.5 package: I see you rebuilt the > flex and bison output files with our current versions of those tools. > I think this is probably okay, but it changes the expected output of > some of the regression tests, eg > > *** ./expected/strings.out Fri Jun 1 13:49:17 2001 > --- ./results/strings.out Thu Aug 19 12:08:22 2004 > *************** > *** 17,23 **** > ' - next line' /* this comment is not allowed here */ > ' - third line' > AS "Illegal comment within continuation"; > ! ERROR: parser: parse error at or near "'" > -- > -- test conversions between various string types > -- > --- 17,23 ---- > ' - next line' /* this comment is not allowed here */ > ' - third line' > AS "Illegal comment within continuation"; > ! ERROR: parser: syntax error at or near "'" > -- > -- test conversions between various string types > -- > > A really paranoid person might complain that we have not adequately > tested that the 7.2 code base will work with the updated tools. > I don't think this is a big deal (I did run through the regression > tests to verify this...) but we will need to fix the regression > expected outputs if we want to do it this way. > > If you want to proceed, let me know and I'll commit the updated > regression outputs into the 7.2 branch, and then we can re-wrap. Let's go with the updated regression outputs ... probably better then going and reverting down flex/bison just for packaging :( ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> I never got any feedback on v7.3.7's packaging, so I'm guessing that it >> looked okay to everyone, even though it was substantially smaller? > > I diff'd the 7.3.6 and 7.3.7 tarballs, and got these unexpected > differences: > > Only in postgresql-7.3.6/contrib/earthdistance: Makefile > Only in postgresql-7.3.6/contrib/earthdistance: README.earthdistance > Only in postgresql-7.3.6/contrib/earthdistance: earthdistance.c > Only in postgresql-7.3.6/contrib/earthdistance: earthdistance.sql.in > Only in postgresql-7.3.6/contrib/earthdistance: expected > Only in postgresql-7.3.6/contrib/earthdistance: sql > > Only in postgresql-7.3.7: postgresql-base-7.3.7.tar.gz.md5 > Only in postgresql-7.3.7: postgresql-docs-7.3.7.tar.gz.md5 > Only in postgresql-7.3.7: postgresql-opt-7.3.7.tar.gz.md5 > Only in postgresql-7.3.7: postgresql-test-7.3.7.tar.gz.md5 > > The lack of earthdistance needs to be fixed (checked out wrong module > no doubt?) The extra md5 files aren't that important but maybe could > be cleaned out when you rebuild. > > Otherwise it looks good, including the docs. Yup, rebuilding 7.3.7 right now ... ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
there, this one looks *much* better ... a couple of k smaller still: %ls -lt postgresql-7.3.7.tar.gz ../v7.3.6/postgresql-7.3.6.tar.gz -rw-r--r-- 1 pgsql pgsql 11284024 Aug 19 16:27 postgresql-7.3.7.tar.gz -rw-r--r-- 1 pgsql pgsql 11288430 Mar 4 21:35 ../v7.3.6/postgresql-7.3.6.tar.gz On Thu, 19 Aug 2004, Marc G. Fournier wrote: > > I've just done a re-export of teh CVS for 7.3.7 ... this time, earthdistance > appears to be in there properly ... not sure what happened, but am building a > new package for it, and will check sizes again when finished ... > > On Thu, 19 Aug 2004, Tom Lane wrote: > >> "Marc G. Fournier" <scrappy@postgresql.org> writes: >>> I never got any feedback on v7.3.7's packaging, so I'm guessing that it >>> looked okay to everyone, even though it was substantially smaller? >> >> Devrim pointed out yesterday that it seemed to be missing >> contrib/earthdistance; that's not enough to explain the size diff >> though. >> >> regards, tom lane >> >> ---------------------------(end of broadcast)--------------------------- >> TIP 8: explain analyze is your friend >> > > ---- > Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) > Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664 > > ---------------------------(end of broadcast)--------------------------- > TIP 8: explain analyze is your friend > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> I just got 7.2.5 done now ... I pulled the man.tar.gz file from 7.2.4 and >> put a copy of it into ~ftp/pub/dev/doc/man-7.2.tar.gz (in case its needed >> in the future) ... > > The 7.2.5 tarball is considerably bigger than 7.2.4; this seems to be > because the built .html files are laying about in doc/src/sgml, instead > of having been deleted again ... 'k, I'm blind ... build scripts for both 7.2 and 7.3 are identical, except for which man.tar.gz it pulls in, and what directories in ftp they get written to ... The section for docs is: cd doc/src gmake postgres.tar.gz mv postgres.tar.gz .. cp /var/spool/ftp/pub/dev/doc/man-7.3.tar.gz ../man.tar.gz cd ../.. gmake split-dist=yes dist gmake maintainer-clean And, looking at the various Makefile(s), I'm not finding anything of significant enough difference insofar as would affect removing the .html files ... Anyone see somethign I'm not seeing? :( ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> Let's go with the updated regression outputs ... probably better then
> going and reverting down flex/bison just for packaging :(
That's what I thought too. Patches committed --- rewrap whenever you
have a chance.
regards, tom lane
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> The section for docs is:
> cd doc/src
> gmake postgres.tar.gz
> mv postgres.tar.gz ..
> cp /var/spool/ftp/pub/dev/doc/man-7.3.tar.gz ../man.tar.gz
> cd ../..
> gmake split-dist=yes dist
> gmake maintainer-clean
Shouldn't there be a "gmake clean" in there before popping back out of
the doc/src subdirectory?
regards, tom lane
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> 'k, I'm blind ... build scripts for both 7.2 and 7.3 are identical, except
> for which man.tar.gz it pulls in, and what directories in ftp they get
> written to ...
Oh, I see the problem. In doc/Makefile, recent versions pass the "make
distclean" operation down to doc/src, but 7.2's doesn't.
It might be safe to alter the Makefile, but I think I'd just vote for
changing your build script instead, seeing that this is probably the
last packaging of 7.2.
Peter, any idea why it was like that back then?
regards, tom lane
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> there, this one looks *much* better ... a couple of k smaller still:
The 7.3.7 tarball looks good to me too. 2 down, 1 to go ...
regards, tom lane
Tom Lane wrote: > Oh, I see the problem. In doc/Makefile, recent versions pass the > "make distclean" operation down to doc/src, but 7.2's doesn't. > Peter, any idea why it was like that back then? The doc tree isn't built by default, so I guess no one had ever thought of cleaning it by default, either. Of course, as it turns out, cleaning it by default is a good idea. -- Peter Eisentraut http://developer.postgresql.org/~petere/
On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> 'k, I'm blind ... build scripts for both 7.2 and 7.3 are identical, except >> for which man.tar.gz it pulls in, and what directories in ftp they get >> written to ... > > Oh, I see the problem. In doc/Makefile, recent versions pass the "make > distclean" operation down to doc/src, but 7.2's doesn't. Done, and done ... sizes look alot better now: svr1# ls -lt postgresql-7.2.5.tar.gz ../v7.2.4/postgresql-7.2.4.tar.gz -rw-r--r-- 1 pgsql pgsql 9273234 Aug 19 20:57 postgresql-7.2.5.tar.gz -rw-r--r-- 1 pgsql pgsql 9244055 Jan 29 2003 ../v7.2.4/postgresql-7.2.4.tar.gz and check the .tar.gz file, and no html files this time ... well, at least none in sgml ... there are some scattered elsewhere though :) ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> Done, and done ... sizes look alot better now:
Almost there, but you didn't pick up my regression test patches :-(
Also, I still see these in the top directory of the full tarfile:
-rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-base-7.2.5.tar.gz.md5
-rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-docs-7.2.5.tar.gz.md5
-rw-r--r-- 1 tgl users 69 Aug 19 19:47 postgresql-opt-7.2.5.tar.gz.md5
-rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-test-7.2.5.tar.gz.md5
Not sure how important that is to fix, but it looks a bit messy.
regards, tom lane
On Thursday 19 August 2004 16:04, Tom Lane wrote:
> "Marc G. Fournier" <scrappy@postgresql.org> writes:
> > Let's go with the updated regression outputs ... probably better then
> > going and reverting down flex/bison just for packaging :(
>
> That's what I thought too. Patches committed --- rewrap whenever you
> have a chance.
>
I may misunderstand the problem, but if the change in bison/flex can break the
regression tests, couldn't it also break peoples apps? That would seem like
a nasty suprise to spring on someone this late in that release cycle...
--
Robert Treat
Build A Better Lamp :: Linux Apache {middleware} PostgreSQL
>> If you want to proceed, let me know and I'll commit the updated >> regression outputs into the 7.2 branch, and then we can re-wrap. > > > Let's go with the updated regression outputs ... probably better then > going and reverting down flex/bison just for packaging :( What about people who are parsing the error messages? Chris
On Thu, 19 Aug 2004, Tom Lane wrote: > "Marc G. Fournier" <scrappy@postgresql.org> writes: >> Done, and done ... sizes look alot better now: > > Almost there, but you didn't pick up my regression test patches :-( Fixed ... > Also, I still see these in the top directory of the full tarfile: > > -rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-base-7.2.5.tar.gz.md5 > -rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-docs-7.2.5.tar.gz.md5 > -rw-r--r-- 1 tgl users 69 Aug 19 19:47 postgresql-opt-7.2.5.tar.gz.md5 > -rw-r--r-- 1 tgl users 70 Aug 19 19:47 postgresql-test-7.2.5.tar.gz.md5 Fixed ... Now, what about Robert/Chris' comments about the potential changes in error messages? Only 3 files were changed as a result of the changes, so it doesn't look to have that big of an impact ... ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
Christopher Kings-Lynne <chriskl@familyhealth.com.au> writes:
> What about people who are parsing the error messages?
Does anyone really have an application that is looking specifically for
"parse error"? And is going to do something useful when it finds it
that it would not do with an unrecognized error message?
Even granting that someone does,
(a) they are at risk anyway, if they choose to build from CVS or the
"make maintainer-clean" tree state on a platform with a recent Bison.
(b) they can always build with an old Bison to get the old behavior.
I was disturbed about the point this morning when I first saw it,
but the more I think about it the more I think it's a non-issue.
regards, tom lane
Hi, On Thu, 19 Aug 2004, Marc G. Fournier wrote: > Devrim/David ... I'm doing a force sync of the ftp site from developer -> > ftp right now, so all the bundles should be there for you ... the SRPMs > are done for 7.4.5, are you going to do similiar with the 7.2.5 and 7.3.7 > distros? Are 7.3.7 and 7.2.5 ready? If so, I want to build them tonight. I'm back from my e-mails now. Regards, -- Devrim GUNDUZ devrim~gunduz.org devrim.gunduz~linux.org.tr http://www.tdmsoft.com http://www.gunduz.org
"Marc G. Fournier" <scrappy@postgresql.org> writes:
> On Thu, 19 Aug 2004, Tom Lane wrote:
>> Almost there, but you didn't pick up my regression test patches :-(
> Fixed ...
Ok, the re-re-wrap looks good AFAICS.
> Now, what about Robert/Chris' comments about the potential changes in
> error messages? Only 3 files were changed as a result of the changes, so
> it doesn't look to have that big of an impact ...
As I said last night, I don't think there's much reason for worry.
regards, tom lane