Subject: [HACKERS] timestamp.c changes
I sent in changes previously and they were rejected because they didn't
follow ANSI spec. Here is the input part of the changes again. Even
though it allows more flexibility for inputting different formats, it
is also backwards compatible with the standard version. I have also
not changed the output format so it will still output the ANSI forms.
Is this acceptable to everyone?
Subject: [HACKERS] More date time functions
Here are some additional patches mostly related to the date and time
data types. It includes some type conversion routines to move between
the different date types and some other date manipulation routines such
as date_part(units,datetime).
I noticed Edmund Mergl et al's neat trick for getting function overloading
for builtin functions, so started to use that for the date and time stuff.
Later, if someone figures out how to get function overloading directly
for internal C code, then we can move to that technique.
These patches include documentation updates (don't faint!) for the built-in
man page. Doesn't yet include mention of timestamp, since I don't know
much about it and since it may change a bit to become a _real_ ANSI timestamp
which would include parser support for the declaration syntax (what do you
think, Dan?).
The patches were developed on the 970330 release, but have been rebuilt
off of the 970402 release. The first patch below is to get libpq to compile,
on my Linux box, but is not related to the rest of the patches and you can
choose not to apply that one at this time. Thanks in advance, scrappy!
Subject: [HACKERS] locale patches !
Hi there,
here are little patches to get Postgres 6.1 works with locale stuff.
This is a patch against 970402.tar.gz, there are no problem to apply them
by hand to 6.0 release. Collate stuff tested about 1-2 months in real
working database but I'm sure there must be no problem. US hackers
could vote against locale implementation ( locale for sure will affect to
speed of postgres ), so I introduce variable USE_LOCALE which
controls locale stuff. Non-US users now could use ~* operator
for searching and <order by> for strings with nation alphabet.
Please, don't forget, as I did first time, to set environment variable
LC_CTYPE and LC_COLLATE because backend get locale information from them.
I start postmaster from a little script, assuming that shell is Bash shell
it looks like:
#!/bin/sh
export LC_CTYPE=koi8-r
export LC_COLLATE=koi8-r
postmaster -B 1024 -S -D/usr/local/pgsql/data/ -o '-Fe'
Subject: [HACKERS] Small date patches (resubmitted)
Here a some small patches for the date/time code. They set the default
output format for the datetime type to the traditional Postgres
style, and fix a date debugging declaration. I submitted these
a couple of days ago, but they might have gotten lost...
NOTE: the second patch to dt.c is what I believe D'Arcy submitted as well,
that I claimed was taken out...sorry D'Arcy, my fault :(
Subject: Re: [HACKERS] abstime "now" broken
Yes, I broke 'now' :( with an attempt at a bug fix involving
servers running in the UTC/GMT timezone. These patches fix
the problem, and have been tested in GMT (+00 hours),
PST (-08), and NZT (+12) timezones which exercized the code for
various cases including across day boundaries. btw, this code
fixes the same type of problem for 'today', 'yesterday', 'tomorrow',
for DATETIME, ABSTIME, DATE and TIME types.
The bugfix itself is quite small, but I have accumulated other
changes in the datetime data type and include them here also.
One set of changes involves printing ISO-formatted dates and
is in response to the helpful information from Kurt Lidl regarding
ANSI SQL dates. I'll send another e-mail sometime soon discussing
more issues he has raised...
Reply-To: hackers@hub.org, Dan McGuirk <mcguirk@indirect.com>
To: hackers@hub.org
Subject: [HACKERS] tmin writeback optimization
I was doing some profiling of the backend, and noticed that during a certain
benchmark I was running somewhere between 30% and 75% of the backend's CPU
time was being spent in calls to TransactionIdDidCommit() from
HeapTupleSatisfiesNow() or HeapTupleSatisfiesItself() to determine that
changed rows' transactions had in fact been committed even though the rows'
tmin values had not yet been set.
When a query looks at a given row, it needs to figure out whether the
transaction that changed the row has been committed and hence it should pay
attention to the row, or whether on the other hand the transaction is still
in progress or has been aborted and hence the row should be ignored. If
a tmin value is set, it is known definitively that the row's transaction
has been committed. However, if tmin is not set, the transaction
referred to in xmin must be looked up in pg_log, and this is what the
backend was spending a lot of time doing during my benchmark.
So, implementing a method suggested by Vadim, I created the following
patch that, the first time a query finds a committed row whose tmin value
is not set, sets it, and marks the buffer where the row is stored as
dirty. (It works for tmax, too.) This doesn't result in the boost in
real time performance I was hoping for, however it does decrease backend
CPU usage by up to two-thirds in certain situations, so it could be
rather beneficial in high-concurrency settings.
Subject: [HACKERS] backend/utils/adt/timestamp.c
Back to this timezone stuff. The struct tm has a field (tm_gmtoff) which
is the offset from UTC (GMT is archaic BTW) in seconds. Is this the
value you are looking for when you use timezone? Note that this applies
to NetBSD but it does not appear to be in either ANSI C or POSIX. This
looks like one of those things that is just going to have to be hand
coded for each platform.
Why not just store the values in UTC and use localtime instead of
gmtime when retrieving the value?
Also, you assume the time is returned as a 4 byte integer. In fact,
there is not even any requirement that time be an integral value. You
should use time_t here.
The input function seems unduly restrictive. Somewhere in the sources
there is an input function that allows words for months. Can't we do
the same here?
There is a standard function, difftime, for subtracting two times. It
deals with cases where time_t is not integral. There is, however, a
small performance hit since it returns a double and I don't believe
there is any system currently which uses anything but an integral for
time_t. Still, this is technically the correct and portable thing to do.
The returns from the various comparisons should probably be a bool.
Subject: [HACKERS] More patches for date/time
I have accumulated several patches to add functionality to the datetime
and timespan data types as well as to fix reported porting bugs on non-BSD
machines. These patches are:
dt.c.patch - add datetime_part(), fix bugs
dt.h.patch - add quarter and timezone support, add prototypes
globals.c.patch - add time and timezone variables
miscadmin.h.patch - add time and timezone variables
nabstime.c.patch - add datetime conversion routine
nabstime.h.patch - add prototypes
pg_operator.h.patch - add datetime operators, clean up formatting
pg_proc.h.patch - add datetime functions, reassign conflicting date OIDs
pg_type.h.patch - add datetime and timespan data types
The dt.c and pg_proc.h patches are fairly large; the latter mostly because I tried
to get some columns for existing entries to line up.
Subject: [HACKERS] backend/utils/adt/nabstime.c
There is a problem with some of the calls to strftime. The second arg is
missing. In all cases the buffer is CTZName which, according to the
file init/globals.c, is char CTZName[8] so I have added this value.
I know there should be a #define set up for this but I wasn't sure
which header to put it in.
According to man page under FreeBSD for sys_errlist[], strerror() should be
used instead...not sure if this will break other systems, so only changing
two files for now, and we'll see what "errors" it turns up
of common routines in pqcomprim.c (pq communication primitives).
Not all adapted to it yet, but it's a start.
- Rewritten some of those routines, to write/read bigger chunks of
data, precomputing stuff in buffers instead of sending out byte
by byte.
- As a consequence, I need to know the endianness of the machine.
Currently I rely on getting it from machine/endian.h, but this
may not be available everywhere? (Who the hell thought it was
a good idea to pass integers to the backend the other way around
than the normal network byte order? *argl*)
- Libpq looks in the environment for magic variables, and upon
establishing a connection to the backend, sends it queries
of the form "SET var_name TO 'var_value'". This needs a change
in the backend parser (Mr. Parser, are you there? :)
- Currently it looks for two Env-Vars, namely PG_DATEFORMAT
and PG_FLOATFORMAT. What else makes sense? PG_TIMEFORMAT?
PG_TIMEZONE?
From: "Martin J. Laubach" <mjl@wwx.vip.at>
Subject: [HACKERS] Patches for 970316 compilation
I made a small pre-emptive change in the new datetime code to eliminate
calls to infnan(). Hopefully this will make Solaris (and probably other
non-GNUlib) systems happier. Didn't find fe-connect.h in the 970316
distribution, so made one up. Also, one of the test routines needs an
update for the geo-decls.h -> geo_decls.h name change.
Patches appear below...
Subject: [HACKERS] linux/alpha patches
These patches lay the groundwork for a Linux/Alpha port. The port doesn't
actually work unless you tweak the linker to put all the pointers in the
first 32 bits of the address space, but it's at least a start. It
implements the test-and-set instruction in Alpha assembly, and also fixes
a lot of pointer-to-integer conversions, which is probably good anyway.
gmake of the code without interruption.
There's also some tidy-up of the MAXPATHLEN stuff based on the assumption that
all supported platforms have MAXPATHLEN defined in <sys/param.h>.
(The only unknowns for the above are AIX and IRIX5.)
And now - JMP_BUF again. Is it enough, folks ?
Fixed again:
cc1: warnings being treated as errors
exc.c: In function 'ExcRaise':
exc.c:187: warning: passing arg 1 of 'Longjmp' from incompatible pointer type
gmake[3]: *** [exc.o] Error 1
cc1: warnings being treated as errors
datum.c: In function `DatumGetSize':
datum.c:57: warning: unsigned value >= 0 is always 1
gmake[3]: *** [datum.o] Error 1
There was:
if (byVal) {
if (len >= 0 && len <= sizeof(Datum)) {
but len has type Size (unsigned int) and so now there is:
if (byVal) {
if (len <= sizeof(Datum)) {
cc1: warnings being treated as errors
exc.c: In function 'ExcRaise':
exc.c:186: warning: passing arg 1 of 'Longjmp' from incompatible pointer type
gmake[3]: *** [exc.o] Error 1
Now we have:
#if defined (JMP_BUF)
longjmp(efp->context, 1);
#else
siglongjmp(efp->context, 1);
#endif
When an acl item is added or updated the new entry is deleted if it has no
permissions and the acl array is shrinked. This is is done by decrementing
the number of items without updating the corresponding array size.
The array with the incorrect size is later read by pg_aclcheck and the entry
count is used to allocate a new array while the array size is used to copy
the old one. This causes a memory corruption and a backend crash.
This happens only to normal user as the administrator bypasses acl checks.
Massimo Dal Zotto
* Wrote max(date) and min(date) aggregates
* Wrote operator "-" for date; date - date yields number of days
difference
* Wrote operator+(date,int) and operator-(date,int); the int is the
number of days. Each operator returns a new date.
By: Tom Tromey <tromey@creche.cygnus.com>
In particular, no more compiled-in default for PGDATA or LIBDIR. Commands
that need them need either invocation options or environment variables.
PGPORT default is hardcoded as 5432, but overrideable with options or
environment variables.
Changes:
* Unique index capability works using the syntax 'create unique
index'.
* Duplicate OID's in the system tables are removed. I put
little scripts called 'duplicate_oids' and 'find_oid' in
include/catalog that help to find and remove duplicate OID's.
I also moved 'unused_oids' from backend/catalog to
include/catalog, since it has to be in the same directory
as the include files in order to work.
* The backend tries converting the name of a function or aggregate
to all lowercase if the original name given doesn't work (mostly
for compatibility with ODBC).
* You can 'SELECT NULL' to your heart's content.
* I put my _bt_updateitem fix in instead, which uses
_bt_insertonpg so that even if the new key is so big that
the page has to be split, everything still works.
* All literal references to system catalog OID's have been
replaced with references to define'd constants from the catalog
header files.
* I added a couple of node copy functions. I think this was a
preliminary attempt to get rules to work.
The comparison routines for text and char data type give incorrect results
if the input data contains characters greater than 127. As these routines
perform the comparison using signed char variables all character codes
greater than 127 are interpreted as less than 0. These codes are used to
encode the iso8859 char sets.
The other text-like data types seem to work as expected as they use unsigned
chars in comparisons.
Submitted by: Massimo Dal Zotto <dz@cs.unitn.it>
conditions are always met. The patch can be applied to any version
of Postgres95 from 1.02 to 1.05. After applying the patch, queries
using indices on bpchar and varchar fields should (hopefully ;-) )
always return the same tuple set regardless to the fact whether
indices are used or not.
Submitted by: Gerhard Reithofer <tbr_laa@AON.AT>
---
below my signature, there are a coupls of diffs and files in a shell
archive, which were needed to build postgres95 1.02 on Siemens Nixdorfs
MIPS based SINIX systems. Except for the compiler switches "-W0" and
"-LD-Blargedynsym" these diffs should also apply for other SVR4 based
systems. The changes in "Makefile.global" and "genbki.sh" can probably
be ignored (I needed gawk, to make the script run).
There is one bugfix thou. In "src/backend/parser/sysfunc.c" the
function in this file didn't honor the EUROPEAN_DATES ifdef.
---
Submitted by: Frank Ridderbusch <ridderbusch.pad@sni.de>
and found out that one of the patches is a show stopper for
compiling under a strict ansi package.
Please make sure the following fix makes it into the 1.02.1
release...
Thanks.
-Kurt
The updating of array fields is broken in Postgres95-1.01, An array can
be only replaced with a new array but not have some elements modified.
This is caused by two bugs in the parser and in the array utilities.
Furthermore it is not possible to update array with a base type of
variable length.
- submitted by: Massimo Dal Zotto <dz@cs.unitn.it>
Select queries with an isnull or notnull clause, like "select * where
somefield isnull", crash the backend if the table has at least one index.
If the indices are deleted the queries work again. Also the explain
command fail in the same way.
The is caused by a bug in subroutine of the optimizer which doesn't check
null values in the clauses.
Submitted by: Massimo Dal Zotto <dz@cs.unitn.it>
This is a patch to prevent an endless loop occuring in the Postgres backend
when a 'warning' error condition generates another warning error contition
in the handler code.
Submitted by: Chris Dunlop, <chris@onthe.net.au>
varchar.diff
------------
This patch was necessary for the OpenLink Postgres Database Agent.
I think this fixes a bug anyway.
The following query demonstrates this bug:
create table foo (bar varchar);
insert into foo values (''); -- no problem
select * from foo where bar = ''; -- fails
causes segmentation fault.
Thanks to: Salvador Ortiz Garcia, Robert Patrick, Paul 'Shag' Walmsley,
and James Cooper for finding and fixing the problem.