Tom Lane [Fri, 16 Oct 2009 22:08:36 +0000 (22:08 +0000)]
Rewrite pam_passwd_conv_proc to be more robust: avoid assuming that the
pam_message array contains exactly one PAM_PROMPT_ECHO_OFF message.
Instead, deal with however many messages there are, and don't throw error
for PAM_ERROR_MSG and PAM_TEXT_INFO messages. This logic is borrowed from
openssh 5.2p1, which hopefully has seen more real-world PAM usage than we
have. Per bug #5121 from Ryan Douglas, which turned out to be caused by
the conv_proc being called with zero messages. Apparently that is normal
behavior given the combination of Linux pam_krb5 with MS Active Directory
as the domain controller.
Patch all the way back, since this code has been essentially untouched
since 7.4. (Surprising we've not heard complaints before.)
Tom Lane [Thu, 15 Oct 2009 23:39:13 +0000 (23:39 +0000)]
First committed version of plpython_unicode_0.out did not actually contain the
required \200 bytes. Let's see if this commit works, or if CVS is messing it up.
Tom Lane [Wed, 14 Oct 2009 22:14:25 +0000 (22:14 +0000)]
Support SQL-compliant triggers on columns, ie fire only if certain columns
are named in the UPDATE's SET list.
Note: the schema of pg_trigger has not actually changed; we've just started
to use a column that was there all along. catversion bumped anyway so that
this commit is included in the history of potentially interesting changes
to system catalog contents.
Rename the new MAX_AUTH_TOKEN_LENGTH #define to PG_MAX_AUTH_MAX_TOKEN_LENGTH,
to make it more obvious that it's a PostgreSQL internal limit, not something
that comes from system header files.
Peter Eisentraut [Wed, 14 Oct 2009 21:59:15 +0000 (21:59 +0000)]
In the configure check for the Python distutils module, use a less obscure
shell construct to hide away the stderr output. Python 3.1 actually core
dumps on the current invocation (http://bugs.python.org/issue7111), but the
new version also has the more general advantage of saving the error message
in config.log for analysis.
Raise the maximum authentication token (Kerberos ticket) size in GSSAPI
and SSPI athentication methods. While the old 2000 byte limit was more than
enough for Unix Kerberos implementations, tickets issued by Windows Domain
Controllers can be much larger.
Tom Lane [Tue, 13 Oct 2009 21:04:01 +0000 (21:04 +0000)]
Add "\pset linestyle ascii/unicode" option to psql, allowing our traditional
ASCII-art style of table output to be upgraded to use Unicode box drawing
characters if desired. By default, psql will use the Unicode characters
whenever client_encoding is UTF8.
The patch forces linestyle=ascii in pg_regress usage, ensuring we don't
break the regression tests in Unicode locales.
Tom Lane [Tue, 13 Oct 2009 14:33:14 +0000 (14:33 +0000)]
Fix ts_stat's failure on empty tsvector.
Also insert a couple of Asserts that check for stack overflow.
Bogus coding appears to be new in 8.4 --- older releases had a much
simpler algorithm here. Per bug #5111.
Tom Lane [Mon, 12 Oct 2009 23:41:45 +0000 (23:41 +0000)]
Use plurals (TABLES, FUNCTIONS, etc) in ALTER DEFAULT PRIVILEGES. We have
the keywords as a consequence of the GRANT ALL patch, so we might as well
use them and make the ALTER commands read more naturally.
Tom Lane [Mon, 12 Oct 2009 18:10:51 +0000 (18:10 +0000)]
Move the handling of SELECT FOR UPDATE locking and rechecking out of
execMain.c and into a new plan node type LockRows. Like the recent change
to put table updating into a ModifyTable plan node, this increases planning
flexibility by allowing the operations to occur below the top level of the
plan tree. It's necessary in any case to restore the previous behavior of
having FOR UPDATE locking occur before ModifyTable does.
This partially refactors EvalPlanQual to allow multiple rows-under-test
to be inserted into the EPQ machinery before starting an EPQ test query.
That isn't sufficient to fix EPQ's general bogosity in the face of plans
that return multiple rows per test row, though. Since this patch is
mostly about getting some plan node infrastructure in place and not about
fixing ten-year-old bugs, I will leave EPQ improvements for another day.
Another behavioral change that we could now think about is doing FOR UPDATE
before LIMIT, but that too seems like it should be treated as a followon
patch.
Tom Lane [Sat, 10 Oct 2009 03:50:15 +0000 (03:50 +0000)]
Improve similar_escape() in two different ways:
* Stop escaping ? and {. As of SQL:2008, SIMILAR TO is defined to have
POSIX-compatible interpretation of ? as well as {m,n} and related constructs,
so we should allow these things through to our regex engine.
* Escape ^ and $. It appears that our regex engine will treat ^^ at the
beginning of the string the same as ^, and similarly for $$ at the end of
the string, which meant that SIMILAR TO was effectively ignoring ^ at the
start of the pattern and $ at the end. Since these are not supposed to be
metacharacters, this is a bug.
The second part of this is arguably a back-patchable bug fix, but I'm
hesitant to do that because it might break applications that are expecting
something like "col SIMILAR TO '^foo$'" to work like a POSIX pattern.
Seems safer to only change it at a major version boundary.
Tom Lane [Sat, 10 Oct 2009 01:43:50 +0000 (01:43 +0000)]
Split the processing of INSERT/UPDATE/DELETE operations out of execMain.c.
They are now handled by a new plan node type called ModifyTable, which is
placed at the top of the plan tree. In itself this change doesn't do much,
except perhaps make the handling of RETURNING lists and inherited UPDATEs a
tad less klugy. But it is necessary preparation for the intended extension of
allowing RETURNING queries inside WITH.
Add a variant of pg_get_triggerdef with a second argument "pretty" that
causes the output to be formatted in the way pg_dump used to do. Use this
variant in pg_dump with server versions >= 8.5.
This insulates pg_dump from most future trigger feature additions, such as
the upcoming column triggers patch.
Tom Lane [Thu, 8 Oct 2009 22:34:57 +0000 (22:34 +0000)]
Remove very ancient tuple-counting infrastructure (IncrRetrieved() and
friends). This code has all been ifdef'd out for many years, and doesn't
seem to have any prospect of becoming any more useful in the future.
EXPLAIN ANALYZE is what people use in practice, and I think if we did want
process-wide counters we'd be more likely to put in dtrace events for that
than try to resurrect this code. Get rid of it so as to have one less detail
to worry about while refactoring execMain.c.
Tom Lane [Thu, 8 Oct 2009 02:39:25 +0000 (02:39 +0000)]
Support use of function argument names to identify which actual arguments
match which function parameters. The syntax uses AS, for example
funcname(value AS arg1, anothervalue AS arg2)
Alvaro Herrera [Wed, 7 Oct 2009 22:14:26 +0000 (22:14 +0000)]
Make it possibly to specify GUC params per user and per database.
Create a new catalog pg_db_role_setting where they are now stored, and better
encapsulate the code that deals with settings into its realm. The old
datconfig and rolconfig columns are removed.
psql has gained a \drds command to display the settings.
Backwards compatibility warning: while the backwards-compatible system views
still have the config columns, they no longer completely represent the
configuration for a user or database.
Alvaro Herrera [Wed, 7 Oct 2009 16:27:18 +0000 (16:27 +0000)]
Fix snapshot management, take two.
Partially revert the previous patch I installed and replace it with a more
general fix: any time a snapshot is pushed as Active, we need to ensure that it
will not be modified in the future. This means that if the same snapshot is
used as CurrentSnapshot, it needs to be copied separately. This affects
serializable transactions only, because CurrentSnapshot has already been copied
by RegisterSnapshot and so PushActiveSnapshot does not think it needs another
copy. However, CommandCounterIncrement would modify CurrentSnapshot, whereas
ActiveSnapshots must not have their command counters incremented.
I say "partially" because the regression test I added for the previous bug
has been kept.
(This restores 8.3 behavior, because before snapmgr.c existed, any snapshot set
as Active was copied.)
Per bug report from Stuart Bishop in
6bc73d4c0910042358k3d1adff3qa36f8df75198ecea@mail.gmail.com
Most things should be cleaned by "make clean", except the parts that are
shipped in the tarball. These rules had gotten a bit out of whack after
the various restructurings of the documentation build rules.
Tom Lane [Tue, 6 Oct 2009 00:55:26 +0000 (00:55 +0000)]
Change CREATE TABLE so that column default expressions coming from different
inheritance parent tables are compared using equal(), instead of doing
strcmp() on the nodeToString representation. The old implementation was
always a tad cheesy, and it finally fails completely as of 8.4, now that the
node tree might contain syntax location information. equal() knows it's
supposed to ignore those fields, but strcmp() hardly can. Per recent
report from Scott Ribe.
Tom Lane [Mon, 5 Oct 2009 19:24:49 +0000 (19:24 +0000)]
Create an ALTER DEFAULT PRIVILEGES command, which allows users to adjust
the privileges that will be applied to subsequently-created objects.
Such adjustments are always per owning role, and can be restricted to objects
created in particular schemas too. A notable benefit is that users can
override the traditional default privilege settings, eg, the PUBLIC EXECUTE
privilege traditionally granted by default for functions.
Tom Lane [Sat, 3 Oct 2009 20:04:39 +0000 (20:04 +0000)]
Fix assorted memory leaks in pg_hba.conf parsing. Over a sufficiently
large number of SIGHUP cycles, these would have run the postmaster out
of memory. Noted while testing memory-leak scenario in postgresql.conf
configuration-change-printing patch.
Tom Lane [Sat, 3 Oct 2009 18:04:57 +0000 (18:04 +0000)]
Fix a couple of issues in recent patch to print updates to postgresql.conf
settings: avoid calling superuser() in contexts where it's not defined,
don't leak the transient copies of GetConfigOption output, and avoid the
whole exercise in postmaster child processes.
I found that actually no current caller of GetConfigOption has any use for
its internal check of GUC_SUPERUSER_ONLY. But rather than just remove
that entirely, it seemed better to add a parameter indicating whether to
enforce the check.
Tom Lane [Fri, 2 Oct 2009 21:14:04 +0000 (21:14 +0000)]
Make sure that GIN fast-insert and regular code paths enforce the same
tuple size limit. Improve the error message for index-tuple-too-large
so that it includes the actual size, the limit, and the index name.
Sync with the btree occurrences of the same error.
Back-patch to 8.4 because it appears that the out-of-sync problem
is occurring in the field.
Tom Lane [Fri, 2 Oct 2009 18:13:04 +0000 (18:13 +0000)]
Fix erroneous handling of shared dependencies (ie dependencies on roles)
in CREATE OR REPLACE FUNCTION. The original code would update pg_shdepend
as if a new function was being created, even if it wasn't, with two bad
consequences: pg_shdepend might record the wrong owner for the function,
and any dependencies for roles mentioned in the function's ACL would be lost.
The fix is very easy: just don't touch pg_shdepend at all when doing a
function replacement.
Also update the CREATE FUNCTION reference page, which never explained
exactly what changes and doesn't change in a function replacement.
In passing, fix the CREATE VIEW reference page similarly; there's no
code bug there, but the docs didn't say what happens.
Alvaro Herrera [Fri, 2 Oct 2009 17:57:30 +0000 (17:57 +0000)]
Ensure that a cursor has an immutable snapshot throughout its lifespan.
The old coding was using a regular snapshot, referenced elsewhere, that was
subject to having its command counter updated. Fix by creating a private copy
of the snapshot exclusively for the cursor.
Backpatch to 8.4, which is when the bug was introduced during the snapshot
management rewrite.
Tom Lane [Wed, 30 Sep 2009 19:50:22 +0000 (19:50 +0000)]
Assorted improvements in contrib/hstore.
Remove the 64K limit on the lengths of keys and values within an hstore.
(This changes the on-disk format, but the old format can still be read.)
Add support for btree/hash opclasses for hstore --- this is not so much
for actual indexing purposes as to allow use of GROUP BY, DISTINCT, etc.
Add various other new functions and operators.
Tom Lane [Tue, 29 Sep 2009 01:20:34 +0000 (01:20 +0000)]
Fix equivclass.c's not-quite-right strategy for handling X=X clauses.
The original coding correctly noted that these aren't just redundancies
(they're effectively X IS NOT NULL, assuming = is strict). However, they
got treated that way if X happened to be in a single-member EquivalenceClass
already, which could happen if there was an ORDER BY X clause, for instance.
The simplest and most reliable solution seems to be to not try to process
such clauses through the EquivalenceClass machinery; just throw them back
for traditional processing. The amount of work that'd be needed to be
smarter than that seems out of proportion to the benefit.
Per bug #5084 from Bernt Marius Johnsen, and analysis by Andrew Gierth.
Andrew Dunstan [Mon, 28 Sep 2009 17:31:12 +0000 (17:31 +0000)]
Convert a perl array to a postgres array when returned by Set Returning Functions as well as non SRFs. Backpatch to 8.1 where these facilities were introduced. with a little help from Abhijit Menon-Sen.
Tom Lane [Sun, 27 Sep 2009 20:09:58 +0000 (20:09 +0000)]
Replace the array-style TupleTable data structure with a simple List of
TupleTableSlot nodes. This eliminates the need to count in advance
how many Slots will be needed, which seems more than worth the small
increase in the amount of palloc traffic during executor startup.
The ExecCountSlots infrastructure is now all dead code, but I'll remove it
in a separate commit for clarity.
Tom Lane [Sun, 27 Sep 2009 03:43:10 +0000 (03:43 +0000)]
Make libpq reject non-numeric and out-of-range port numbers with a suitable
error message, rather than blundering on and failing with something opaque.
Tom Lane [Sun, 27 Sep 2009 01:32:11 +0000 (01:32 +0000)]
Simplify the bootstrap (BKI) code by getting rid of a useless table of all
the strings seen during the bootstrap run. There might have been some
actual point to doing that, many years ago, but as far as I can see the only
value now is to conserve a bit of memory. Even if we cared about wasting
a megabyte or so during the initdb run, it'd be far more effective to
arrange to release memory at the end of each BKI command, instead of
intentionally hanging onto strings that might never be used again.
Not maintaining the table probably makes it faster too; but the main point
of this patch is to get rid of a couple hundred lines of unnecessary and
rather crufty code.
Tom Lane [Sat, 26 Sep 2009 23:08:22 +0000 (23:08 +0000)]
Revert my ill-considered change that made formrdesc not insert the correct
relation rowtype OID into the relcache entries it builds. This ensures
that catcache copies of the relation tupdescs will be fully correct.
While the deficiency doesn't seem to have any effect in the current
sources, we have been bitten by not-quite-right catcache tupdescs before,
so it seems like a good idea to maintain the rule that they should be right.
Tom Lane [Sat, 26 Sep 2009 22:42:03 +0000 (22:42 +0000)]
Extend the BKI infrastructure to allow system catalogs to be given
hand-assigned rowtype OIDs, even when they are not "bootstrapped" catalogs
that have handmade type rows in pg_type.h. Give pg_database such an OID.
Restore the availability of C macros for the rowtype OIDs of the bootstrapped
catalogs. (These macros are now in the individual catalogs' .h files,
though, not in pg_type.h.)
This commit doesn't do anything especially useful by itself, but it's
necessary infrastructure for reverting some ill-considered changes in
relcache.c.
Tom Lane [Sat, 26 Sep 2009 18:24:49 +0000 (18:24 +0000)]
Fix RelationCacheInitializePhase2 (Phase3, in HEAD) to cope with the
possibility of shared-inval messages causing a relcache flush while it tries
to fill in missing data in preloaded relcache entries. There are actually
two distinct failure modes here:
1. The flush could delete the next-to-be-processed cache entry, causing
the subsequent hash_seq_search calls to go off into the weeds. This is
the problem reported by Michael Brown, and I believe it also accounts
for bug #5074. The simplest fix is to restart the hashtable scan after
we've read any new data from the catalogs. It appears that pre-8.4
branches have not suffered from this failure, because by chance there were
no other catalogs sharing the same hash chains with the catalogs that
RelationCacheInitializePhase2 had work to do for. However that's obviously
pretty fragile, and it seems possible that derivative versions with
additional system catalogs might be vulnerable, so I'm back-patching this
part of the fix anyway.
2. The flush could delete the *current* cache entry, in which case the
pointer to the newly-loaded data would end up being stored into an
already-deleted Relation struct. As long as it was still deleted, the only
consequence would be some leaked space in CacheMemoryContext. But it seems
possible that the Relation struct could already have been recycled, in
which case this represents a hard-to-reproduce clobber of cached data
structures, with unforeseeable consequences. The fix here is to pin the
entry while we work on it.
In passing, also change RelationCacheInitializePhase2 to Assert that
formrdesc() set up the relation's cached TupleDesc (rd_att) with the
correct type OID and hasoids values. This is more appropriate than
silently updating the values, because the original tupdesc might already
have been copied into the catcache. However this part of the patch is
not in HEAD because it fails due to some questionable recent changes in
formrdesc :-(. That will be cleaned up in a subsequent patch.
Tom Lane [Tue, 22 Sep 2009 23:43:43 +0000 (23:43 +0000)]
Implement the DO statement to support execution of PL code without having
to create a function for it.
Procedural languages now have an additional entry point, namely a function
to execute an inline code block. This seemed a better design than trying
to hide the transient-ness of the code from the PL. As of this patch, only
plpgsql has an inline handler, but probably people will soon write handlers
for the other standard PLs.
In passing, remove the long-dead LANCOMPILER option of CREATE LANGUAGE.
Tom Lane [Mon, 21 Sep 2009 20:10:21 +0000 (20:10 +0000)]
Define a new, more extensible syntax for COPY options.
This is intentionally similar to the recently revised syntax for EXPLAIN
options, ie, (name value, ...). The old syntax is still supported for
backwards compatibility, but we intend that any options added in future
will be provided only in the new syntax.
Tom Lane [Sun, 20 Sep 2009 01:53:32 +0000 (01:53 +0000)]
Allow plpgsql IN parameters to be assigned to. Since the parameters are just
preinitialized local variables, this does not affect the function's semantics
as seen by callers; allowing assignment simply avoids the need to create more
local variables in some cases. In any case we were being rather inconsistent
since only scalar parameters were getting marked constant.
No documentation change, since parameters were never documented as being
marked constant anyway.
Tom Lane [Sat, 19 Sep 2009 21:51:21 +0000 (21:51 +0000)]
Remove a couple hundred lines of ugly and tedious-to-maintain code by not
trying to parse COPY options exactly in psql's \copy support. Instead,
just send the options as-is and let the backend sort it out.
Tom Lane [Fri, 18 Sep 2009 17:24:51 +0000 (17:24 +0000)]
Marginal code cleanup in joinpath.c: factor out clause variable-membership
tests into a small common subroutine, and eliminate an unnecessary difference
in the order in which conditions are tested. Per a comment from Robert Haas.
Peter Eisentraut [Fri, 18 Sep 2009 05:00:42 +0000 (05:00 +0000)]
Easier to translate psql help
Instead of requiring translators to translate the entire SQL command
synopses, change create_help.pl to only require them to translate the
placeholders, and paste those into the synopsis using a printf mechanism.
Make some small updates to the markup to make it easier to parse.
Note: This causes msgmerge of gettext 0.17 to segfault. You will need
the patch from https://savannah.gnu.org/bugs/?27474 to make it work.
msgmerge usually only runs on babel.postgresql.org, however.
Peter Eisentraut [Thu, 17 Sep 2009 21:15:18 +0000 (21:15 +0000)]
Improve wording of error message when a postgresql.conf setting is
ignored because it can only be set at server start. In particular,
hiding the main reason in the detail message was suboptimal.
Tom Lane [Thu, 17 Sep 2009 20:49:29 +0000 (20:49 +0000)]
Implement "join removal" for cases where the inner side of a left join
is unique and is not referenced above the join. In this case the inner
side doesn't affect the query result and can be thrown away entirely.
Although perhaps nobody would ever write such a thing by hand, it's
a reasonably common case in machine-generated SQL.
The current implementation only recognizes the case where the inner side
is a simple relation with a unique index matching the query conditions.
This is enough for the use-cases that have been shown so far, but we
might want to try to handle other cases later.
Tom Lane [Tue, 15 Sep 2009 20:31:30 +0000 (20:31 +0000)]
Fix two distinct errors in creation of GIN_INSERT_LISTPAGE xlog records.
In practice these mistakes were always masked when full_page_writes was on,
because XLogInsert would always choose to log the full page, and then
ginRedoInsertListPage wouldn't try to do anything. But with full_page_writes
off a WAL replay failure was certain.
The GIN_INSERT_LISTPAGE record type could probably be eliminated entirely
in favor of using XLOG_HEAP_NEWPAGE, but I refrained from doing that now
since it would have required a significantly more invasive patch.
In passing do a little bit of code cleanup, including making the accounting
for free space on GIN list pages more precise. (This wasn't a bug as the
errors were always in the conservative direction.)
Per report from Simon. Back-patch to 8.4 which contains the identical code.
Tom Lane [Tue, 15 Sep 2009 02:31:15 +0000 (02:31 +0000)]
Fix possible buffer overrun and/or unportable behavior in pg_md5_encrypt()
if salt_len == 0. This seems to be mostly academic, since nearly all calling
code paths guarantee nonempty salt; the only case that doesn't is
PQencryptPassword where the caller could mistakenly pass an empty username.
So, fix it but don't bother backpatching. Per ljb.
Tom Lane [Sun, 13 Sep 2009 22:18:22 +0000 (22:18 +0000)]
Write psql's ~/.psql_history file using history_truncate_file() and
append_history(), if libreadline is new enough to have those functions
(they seem to be present at least since 4.2; but libedit may not have them).
This gives significantly saner behavior when two or more sessions overlap in
their use of the history file; although having two sessions exit at just the
same time is still perilous to your history. The behavior of \s remains
unchanged, ie, overwrite whatever was there.
Per bug #5052 from Marek Wójtowicz.
Peter Eisentraut [Sun, 13 Sep 2009 22:07:06 +0000 (22:07 +0000)]
Fix Unicode support in PL/Python
Check calls of PyUnicode_AsEncodedString() for NULL return, probably
because the encoding name is not known. Add special treatment for
SQL_ASCII, which Python definitely does not know.
Since using SQL_ASCII produces errors in the regression tests when
non-ASCII characters are involved, we have to put back various regression
test result variants.
Don't error out if recycling or removing an old WAL segment fails at the end
of checkpoint. Although the checkpoint has been written to WAL at that point
already, so that all data is safe, and we'll retry removing the WAL segment at
the next checkpoint, if such a failure persists we won't be able to remove any
other old WAL segments either and will eventually run out of disk space. It's
better to treat the failure as non-fatal, and move on to clean any other WAL
segment and continue with any other end-of-checkpoint cleanup.
We don't normally expect any such failures, but on Windows it can happen with
some anti-virus or backup software that lock files without FILE_SHARE_DELETE
flag.
Also, the loop in pgrename() to retry when the file is locked was broken. If a
file is locked on Windows, you get ERROR_SHARE_VIOLATION, not
ERROR_ACCESS_DENIED, at least on modern versions. Fix that, although I left
the check for ERROR_ACCESS_DENIED in there as well (presumably it was correct
in some environment), and added ERROR_LOCK_VIOLATION to be consistent with
similar checks in pgwin32_open(). Reduce the timeout on the loop from 30s to
10s, on the grounds that since it's been broken, we've effectively had a
timeout of 0s and no-one has complained, so a smaller timeout is actually
closer to the old behavior. A longer timeout would mean that if recycling a
WAL file fails because it's locked for some reason, InstallXLogFileSegment()
will hold ControlFileLock for longer, potentially blocking other backends, so
a long timeout isn't totally harmless.
While we're at it, set errno correctly in pgrename().
Backpatch to 8.2, which is the oldest version supported on Windows. The xlog.c
changes would make sense on other platforms and thus on older versions as
well, but since there's no such locking issues on other platforms, it's not
worth it.
Joe Conway [Sat, 12 Sep 2009 23:20:52 +0000 (23:20 +0000)]
plug dblink resource leak
dblink generates orphaned connections when called with a connection string,
fail_on_error = true, and an ERROR occurs. Discovery and patch by
Tatsuhito Kasahara. Introduced in 8.4.
Peter Eisentraut [Sat, 12 Sep 2009 22:13:12 +0000 (22:13 +0000)]
Add Unicode support in PL/Python
PL/Python now accepts Unicode objects where it previously only accepted string
objects (for example, as return value). Unicode objects are converted to the
PostgreSQL server encoding as necessary.
This change is also necessary for future Python 3 support, which treats all
strings as Unicode objects.
Since this removes the error conditions that the plpython_unicode test file
tested for, the alternative result files are no longer necessary.
Tom Lane [Sat, 12 Sep 2009 22:12:09 +0000 (22:12 +0000)]
Rewrite the planner's handling of materialized plan types so that there is
an explicit model of rescan costs being different from first-time costs.
The costing of Material nodes in particular now has some visible relationship
to the actual runtime behavior, where before it was essentially fantasy.
This also fixes up a couple of places where different materialized plan types
were treated differently for no very good reason (probably just oversights).
A couple of the regression tests are affected, because the planner now chooses
to put the other relation on the inside of a nestloop-with-materialize.
So far as I can see both changes are sane, and the planner is now more
consistently following the expectation that it should prefer to materialize
the smaller of two relations.