6.5.1

On Fri, 9 Jul 1999, Bruce Momjian wrote:

Date: Fri, 9 Jul 1999 14:00:10 -0400 (EDT)
From: Bruce Momjian <maillist@candle.pha.pa.us>
To: PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: [HACKERS] 6.5.1

[Man, I am applying all the fixes. If there is a problem with 6.5.1,
they are going to know it was me.]

Just updated cvs and source dosn't compiled on my Linux box 2.0.37
using egcs 1.12 release:

make[2]: Entering directory /home/postgres/cvs/pgsql/src/backend/access'
make -C common SUBSYS.o
make[3]: Entering directory /home/postgres/cvs/pgsql/src/backend/access/common'
gcc -I../../../include -I../../../backend -O2 -mpentium -Wall -Wmissing-prototypes -I../.. -c heaptuple.c -o heaptuple.o
In file included from heaptuple.c:22:
../../../include/access/heapam.h:30: parse error before ime_t'
../../../include/access/heapam.h:30: warning: no semicolon at end of struct or union
../../../include/access/heapam.h:31: warning: type defaults to nt' in declaration of ocal_reset_timestamp'
../../../include/access/heapam.h:31: warning: data definition has no type or storage class
../../../include/access/heapam.h:32: parse error before ast_request_timestamp'
../../../include/access/heapam.h:32: warning: type defaults to nt' in declaration of ast_request_timestamp'
../../../include/access/heapam.h:32: warning: data definition has no type or storage class
../../../include/access/heapam.h:79: parse error before }'
../../../include/access/heapam.h:79: warning: type defaults to nt' in declaration of eapAccessStatisticsData'
../../../include/access/heapam.h:79: warning: data definition has no type or storage class
../../../include/access/heapam.h:81: parse error before *'
../../../include/access/heapam.h:81: warning: type defaults to nt' in declaration of eapAccessStatistics'
../../../include/access/heapam.h:81: warning: data definition has no type or storage class
../../../include/access/heapam.h:238: parse error before eap_access_stats'
../../../include/access/heapam.h:238: warning: type defaults to nt' in declaration of eap_access_stats'
../../../include/access/heapam.h:238: warning: data definition has no type or storage class
../../../include/access/heapam.h:284: parse error before tats'
make[3]: *** [heaptuple.o] Error 1

Regards,
Oleg

-- 
Bruce Momjian                        |  http://www.op.net/~candle
maillist@candle.pha.pa.us            |  (610) 853-3000
+  If your life is a hard drive,     |  830 Blythe Avenue
+  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

On Fri, 9 Jul 1999, Bruce Momjian wrote:

Date: Fri, 9 Jul 1999 14:00:10 -0400 (EDT)
From: Bruce Momjian <maillist@candle.pha.pa.us>
To: PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: [HACKERS] 6.5.1

[Man, I am applying all the fixes. If there is a problem with 6.5.1,
they are going to know it was me.]

Just updated cvs and source dosn't compiled on my Linux box 2.0.37
using egcs 1.12 release:

make[2]: Entering directory /home/postgres/cvs/pgsql/src/backend/access'
make -C common SUBSYS.o
make[3]: Entering directory /home/postgres/cvs/pgsql/src/backend/access/common'
gcc -I../../../include -I../../../backend -O2 -mpentium -Wall -Wmissing-prototypes -I../.. -c heaptuple.c -o heaptuple.o
In file included from heaptuple.c:22:
../../../include/access/heapam.h:30: parse error before ime_t'
../../../include/access/heapam.h:30: warning: no semicolon at end of struct or union
../../../include/access/heapam.h:31: warning: type defaults to nt' in declaration of ocal_reset_timestamp'
../../../include/access/heapam.h:31: warning: data definition has no type or storage class
../../../include/access/heapam.h:32: parse error before ast_request_timestamp'
../../../include/access/heapam.h:32: warning: type defaults to nt' in declaration of ast_request_timestamp'
../../../include/access/heapam.h:32: warning: data definition has no type or storage class
../../../include/access/heapam.h:79: parse error before }'
../../../include/access/heapam.h:79: warning: type defaults to nt' in declaration of eapAccessStatisticsData'
../../../include/access/heapam.h:79: warning: data definition has no type or storage class
../../../include/access/heapam.h:81: parse error before *'
../../../include/access/heapam.h:81: warning: type defaults to nt' in declaration of eapAccessStatistics'
../../../include/access/heapam.h:81: warning: data definition has no type or storage class
../../../include/access/heapam.h:238: parse error before eap_access_stats'
../../../include/access/heapam.h:238: warning: type defaults to nt' in declaration of eap_access_stats'
../../../include/access/heapam.h:238: warning: data definition has no type or storage class
../../../include/access/heapam.h:284: parse error before tats'
make[3]: *** [heaptuple.o] Error 1

Oleg, you are not going to catch me this easily. :-)

Looks like a problem on your end. Try removing heapam.h and re-cvs'ing
it.

I just did a:

#$ rm heapam.h
#$ pgcvs update heapam.h

and line 30 looks fine:

time_t init_global_timestamp; /* time global statistics started */
time_t local_reset_timestamp; /* last time local reset was done */

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Don't worry about it Bruce, we'd blame you anyway. :)
-DEJ

yOn Fri, 9 Jul 1999, Bruce Momjian wrote:

Date: Fri, 9 Jul 1999 16:10:19 -0400 (EDT)
From: Bruce Momjian <maillist@candle.pha.pa.us>
To: Oleg Bartunov <oleg@sai.msu.su>
Cc: PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: Re: [HACKERS] 6.5.1

Oleg, you are not going to catch me this easily. :-)

Looks like a problem on your end. Try removing heapam.h and re-cvs'ing
it.

I just did a:

#$ rm heapam.h
#$ pgcvs update heapam.h

and line 30 looks fine:

time_t init_global_timestamp; /* time global statistics started */
time_t local_reset_timestamp; /* last time local reset was done */

Hmm, removed heapam.h, resynced source, checked heapam.h, see nothing wrong.
But the problem persists :-(
Will try FreeBSD, the same problem . Compiler is the same: egsc 1.12 release
6.5 was compiled fine on both platforms

Regards,
Oleg

-- 
Bruce Momjian                        |  http://www.op.net/~candle
maillist@candle.pha.pa.us            |  (610) 853-3000
+  If your life is a hard drive,     |  830 Blythe Avenue
+  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

Don't worry about it Bruce, we'd blame you anyway. :)
-DEJ

What a group of guys. :-)

-----Original Message-----
From: Bruce Momjian [SMTP:maillist@candle.pha.pa.us]
Sent: Friday, July 09, 1999 1:00 PM
To: PostgreSQL-development
Subject: [HACKERS] 6.5.1

[Man, I am applying all the fixes. If there is a problem with 6.5.1,
they are going to know it was me.]

-- 
Bruce Momjian                        |  http://www.op.net/~candle
maillist@candle.pha.pa.us            |  (610) 853-3000
+  If your life is a hard drive,     |  830 Blythe Avenue
+  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

I just did a:

#$ rm heapam.h
#$ pgcvs update heapam.h

and line 30 looks fine:

time_t init_global_timestamp; /* time global statistics started */
time_t local_reset_timestamp; /* last time local reset was done */

Hmm, removed heapam.h, resynced source, checked heapam.h, see nothing wrong.
But the problem persists :-(
Will try FreeBSD, the same problem . Compiler is the same: egsc 1.12 release
6.5 was compiled fine on both platforms

OK, what do you see in those files. Do you see a ^T there? I have just
recompiled everything here, and have no problems. Strange.

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Hmm, removed heapam.h, resynced source, checked heapam.h, see nothing wrong.
But the problem persists :-(
Will try FreeBSD, the same problem . Compiler is the same: egsc 1.12 release
6.5 was compiled fine on both platforms

OK, what do you see in those files. Do you see a ^T there? I have just
recompiled everything here, and have no problems. Strange.

Honest, guys, I didn't introduce this bug. Honest. No, Marc, noooo...
:-)

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

On Fri, 9 Jul 1999, Bruce Momjian wrote:

Date: Fri, 9 Jul 1999 17:15:49 -0400 (EDT)
From: Bruce Momjian <maillist@candle.pha.pa.us>
To: Oleg Bartunov <oleg@sai.msu.su>
Cc: PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: Re: [HACKERS] 6.5.1

I just did a:

#$ rm heapam.h
#$ pgcvs update heapam.h

and line 30 looks fine:

time_t init_global_timestamp; /* time global statistics started */
time_t local_reset_timestamp; /* last time local reset was done */

Hmm, removed heapam.h, resynced source, checked heapam.h, see nothing wrong.
But the problem persists :-(
Will try FreeBSD, the same problem . Compiler is the same: egsc 1.12 release
6.5 was compiled fine on both platforms

OK, what do you see in those files. Do you see a ^T there? I have just
recompiled everything here, and have no problems. Strange.

Nothing strange in heapam.h I just tried standard gcc 2.7.2.3 and
ended with the same problem. Probably configure problem on Linux system

Oleg

-- 
Bruce Momjian                        |  http://www.op.net/~candle
maillist@candle.pha.pa.us            |  (610) 853-3000
+  If your life is a hard drive,     |  830 Blythe Avenue
+  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

Oleg Bartunov <oleg@sai.msu.su> writes:

Nothing strange in heapam.h I just tried standard gcc 2.7.2.3 and
ended with the same problem. Probably configure problem on Linux system

I'm seeing no problem with cvs sources from yesterday evening. I think
you must have a corrupted copy of one of the files --- not heapam.h,
evidently, but maybe something it depends on. Try removing and
refetching everything that was pulled by your last cvs run.

This isn't the first time we've seen this sort of report. Perhaps
'cvs update' is subject to file damage over a flaky connection?
I wonder if there's anything we can do to increase its reliability.

regards, tom lane

On Sat, 10 Jul 1999, Tom Lane wrote:

Date: Sat, 10 Jul 1999 11:06:51 -0400
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Oleg Bartunov <oleg@sai.msu.su>
Cc: Bruce Momjian <maillist@candle.pha.pa.us>,
PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: Re: [HACKERS] 6.5.1

Oleg Bartunov <oleg@sai.msu.su> writes:

Nothing strange in heapam.h I just tried standard gcc 2.7.2.3 and
ended with the same problem. Probably configure problem on Linux system

I'm seeing no problem with cvs sources from yesterday evening. I think
you must have a corrupted copy of one of the files --- not heapam.h,
evidently, but maybe something it depends on. Try removing and
refetching everything that was pulled by your last cvs run.

Still no luck :-( I did fresh cvs checkout.

This isn't the first time we've seen this sort of report. Perhaps
'cvs update' is subject to file damage over a flaky connection?
I wonder if there's anything we can do to increase its reliability.

regards, tom lane

_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

Oleg Bartunov <oleg@sai.msu.su> writes:

I'm seeing no problem with cvs sources from yesterday evening. I think
you must have a corrupted copy of one of the files --- not heapam.h,
evidently, but maybe something it depends on. Try removing and
refetching everything that was pulled by your last cvs run.

Still no luck :-( I did fresh cvs checkout.

I just did one too, and diffed it against what I had before.
There's still nothing that looks broken.

After looking again at your message, I wonder whether the rest of us
are chasing the wrong idea. The message you sent looked to be corrupted
text, because it mentioned '^Time_t' and so forth. But now I wonder
whether that error wasn't just in your cutting and pasting of the
error message. If we take the messages at face value they seem to
indicate that type time_t is not known to the compiler when it processes
heapam.h, which would make sense if <time.h> hasn't been included yet.

And, right offhand, I'm not seeing where <time.h> gets included before
heapam.h is read.

Has anyone changed anything that might affect where <time.h> gets
included? Perhaps this is a configuration problem.

Oleg, how long ago did you last pull a working fileset?

regards, tom lane

I wrote:

And, right offhand, I'm not seeing where <time.h> gets included before
heapam.h is read.

I dug into this and found that on my own machine, <sys/time.h> is pulled
in by <arpa/inet.h> which is pulled in by config.h (if the right
configuration symbols are defined). It looks to me like there is
noplace that explicitly pulls in <time.h> before heapam.h is read.

In short, what we've got here is code that only works because of
interdependencies among system headers. Not too portable.

I added "#include <time.h>" to heapam.h, which I think will fix Oleg's
problem, but I'm a little bit mystified why we didn't find this long
ago. Someone must have removed an #include somewhere that covered up
the problem...

regards, tom lane

One thing I want to do for 6.6 is make sure each include file has the
proper includes to compile just itself, and try to remove extra
non-system includes in the C files.

I wrote:

And, right offhand, I'm not seeing where <time.h> gets included before
heapam.h is read.

I dug into this and found that on my own machine, <sys/time.h> is pulled
in by <arpa/inet.h> which is pulled in by config.h (if the right
configuration symbols are defined). It looks to me like there is
noplace that explicitly pulls in <time.h> before heapam.h is read.

In short, what we've got here is code that only works because of
interdependencies among system headers. Not too portable.

I added "#include <time.h>" to heapam.h, which I think will fix Oleg's
problem, but I'm a little bit mystified why we didn't find this long
ago. Someone must have removed an #include somewhere that covered up
the problem...

regards, tom lane

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

One thing I want to do for 6.6 is make sure each include file has the
proper includes to compile just itself, and try to remove extra
non-system includes in the C files.

I realize not to touch the system includes, because just because my OS
doesn't need it, doesn't mean others don't.

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

On Sat, 10 Jul 1999, Tom Lane wrote:

Date: Sat, 10 Jul 1999 17:51:00 -0400
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Oleg Bartunov <oleg@sai.msu.su>
Cc: Bruce Momjian <maillist@candle.pha.pa.us>,
PostgreSQL-development <pgsql-hackers@postgreSQL.org>
Subject: Re: [HACKERS] 6.5.1

Oleg Bartunov <oleg@sai.msu.su> writes:

I'm seeing no problem with cvs sources from yesterday evening. I think
you must have a corrupted copy of one of the files --- not heapam.h,
evidently, but maybe something it depends on. Try removing and
refetching everything that was pulled by your last cvs run.

Still no luck :-( I did fresh cvs checkout.

I just did one too, and diffed it against what I had before.
There's still nothing that looks broken.

After looking again at your message, I wonder whether the rest of us
are chasing the wrong idea. The message you sent looked to be corrupted
text, because it mentioned '^Time_t' and so forth. But now I wonder
whether that error wasn't just in your cutting and pasting of the
error message. If we take the messages at face value they seem to

Ooh, soorry. This is known problem (at least for me) with cut-n-paste
in xterm !

indicate that type time_t is not known to the compiler when it processes
heapam.h, which would make sense if <time.h> hasn't been included yet.

Sure, something is broken in configure

And, right offhand, I'm not seeing where <time.h> gets included before
heapam.h is read.

Has anyone changed anything that might affect where <time.h> gets
included? Perhaps this is a configuration problem.

Oleg, how long ago did you last pull a working fileset?

First time I noticed the problem was about 2 weeks ago.

regards, tom lane

_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

I just did one too, and diffed it against what I had before.
There's still nothing that looks broken.

After looking again at your message, I wonder whether the rest of us
are chasing the wrong idea. The message you sent looked to be corrupted
text, because it mentioned '^Time_t' and so forth. But now I wonder
whether that error wasn't just in your cutting and pasting of the
error message. If we take the messages at face value they seem to
indicate that type time_t is not known to the compiler when it processes
heapam.h, which would make sense if <time.h> hasn't been included yet.

And, right offhand, I'm not seeing where <time.h> gets included before
heapam.h is read.

Has anyone changed anything that might affect where <time.h> gets
included? Perhaps this is a configuration problem.

Oleg, how long ago did you last pull a working fileset?

OK, I think I have found the cause. At one point, include/access/htup.h
had included #include <utils/nabstime.h>, though it did not need that
file. I removed the include because it was not needed by that file.
Now, it turns out other files needed it. I am putting it back in, and
this will all be cleaned up by 6.6.

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026