|
| 1 | +2001-05-18 12:00 tgl |
| 2 | + |
| 3 | +* src/include/catalog/pg_aggregate.h (1.29.2.1): Correct |
| 4 | +recently-broken avg(interval) definition. We can't force an initdb |
| 5 | +to fix this in 7.1 installations, but it seems better to be |
| 6 | +shipping a correct entry than a wrong one. |
| 7 | + |
| 8 | +2001-05-17 09:29 momjian |
| 9 | + |
| 10 | +* HISTORY (1.106.2.2), doc/src/sgml/release.sgml (1.90.2.2): Add |
| 11 | +HISTORY items for 7.1.2. |
| 12 | + |
| 13 | +2001-05-16 20:48 tgl |
| 14 | + |
| 15 | +* src/backend/access/heap/heapam.c (1.113.2.1): Back-patch fix for |
| 16 | +race condition in heap_update (make sure we hold the buffer lock |
| 17 | +while checking page free space). |
| 18 | + |
| 19 | +2001-05-15 13:05 momjian |
| 20 | + |
| 21 | +* src/backend/access/gist/: gist.c (1.72.2.1), gistget.c |
| 22 | +(1.25.2.1): Update crypto with patch for computation. |
| 23 | + |
| 24 | +2001-05-15 00:45 momjian |
| 25 | + |
| 26 | +* contrib/pgcrypto/encode.c (1.4.2.2): Well, the correct code - |
| 27 | +that corresponds to current encode - is below.I even got the |
| 28 | +linefeed stuff wrong. |
| 29 | + |
| 30 | +-- marko |
| 31 | + |
| 32 | +2001-05-14 20:34 tgl |
| 33 | + |
| 34 | +* src/: backend/executor/execMain.c (1.139.2.1), |
| 35 | +backend/executor/nodeAppend.c (1.40.2.2), include/nodes/execnodes.h |
| 36 | +(1.57.2.1): EvalPlanQual was thoroughly broken for concurrent |
| 37 | +update/delete on inheritance trees (mostly my fault). Repair. |
| 38 | +Also fix long-standing bug in ExecReplace: after recomputing a |
| 39 | +concurrently updated tuple, we must recheck constraints. Make |
| 40 | +EvalPlanQual leak memory with somewhat less enthusiasm than before, |
| 41 | +although plugging leaks fully will require more changes than I care |
| 42 | +to risk in a dot-release. |
| 43 | + |
| 44 | +2001-05-14 16:25 tgl |
| 45 | + |
| 46 | +* src/backend/optimizer/plan/initsplan.c (1.59.2.1): Current |
| 47 | +implementation of FOR UPDATE has no hope of working correctly for |
| 48 | +relations on the nullable side of an OUTER JOIN. For now I think |
| 49 | +we'd better refuse such queries. |
| 50 | + |
| 51 | +2001-05-14 11:11 momjian |
| 52 | + |
| 53 | +* doc/src/sgml/installation.sgml (1.45.2.1): Fix 'set path' for |
| 54 | +csh, path -> $path. |
| 55 | + |
| 56 | +Pierce Tyler |
| 57 | + |
| 58 | +2001-05-13 07:36 petere |
| 59 | + |
| 60 | +* src/bin/pg_config/pg_config.sh (1.2.2.1): Cope with configure |
| 61 | +arguments that contain spaces. |
| 62 | + |
| 63 | +2001-05-12 23:27 momjian |
| 64 | + |
| 65 | +* contrib/pgcrypto/encode.c (1.4.2.1): update. |
| 66 | + |
| 67 | +2001-05-12 19:36 tgl |
| 68 | + |
| 69 | +* src/bin/pg_dump/pg_dump.c (1.205.2.2): proisstrict must be |
| 70 | +assumed FALSE when dumping from a 7.0 database, not TRUE. |
| 71 | +Otherwise we break pl call handler functions. fmgr_oldstyle will |
| 72 | +take care of making sure the semantics are the same for C |
| 73 | +functions. Clean up some slightly grotty coding in 7.0 pg_class |
| 74 | +reading, also. |
| 75 | + |
| 76 | +2001-05-12 16:45 momjian |
| 77 | + |
| 78 | +* src/interfaces/jdbc/org/postgresql/jdbc1/Statement.java |
| 79 | +(1.7.2.1): Backpatch jdbc1 compile failure fix. |
| 80 | + |
| 81 | +2001-05-11 21:09 pjw |
| 82 | + |
| 83 | +* src/bin/pg_dump/: pg_backup_archiver.c (1.25.2.1), |
| 84 | +pg_backup_archiver.h (1.32.2.1), pg_dump.c (1.205.2.1): - Don't |
| 85 | +dump COMMENTs in data-only dumps |
| 86 | + - Fix view dumping SQL for V7.0 |
| 87 | + - Fix bug when getting view oid with long view names |
| 88 | + - Treat SEQUENCE SET TOC entries as data entries rather than |
| 89 | +schema |
| 90 | + entries. |
| 91 | + - Make allowance for data entries that did not have a data dumper |
| 92 | + routine (eg. SEQUENCE SET) |
| 93 | + |
| 94 | +2001-05-11 18:26 petere |
| 95 | + |
| 96 | +* doc/FAQ_MSWIN (1.9.2.1): no more Andy Piper tools |
| 97 | + |
| 98 | +2001-05-10 21:35 momjian |
| 99 | + |
| 100 | +* configure (1.127.2.1): Can't run autoconf on rel, so just |
| 101 | +manually update. |
| 102 | + |
| 103 | +2001-05-10 21:34 momjian |
| 104 | + |
| 105 | +* HISTORY (1.106.2.1), README (1.23.2.1), configure.in (1.120.2.1), |
| 106 | +register.txt (1.31.2.1), doc/FAQ (1.104.2.1), doc/bug.template |
| 107 | +(1.18.2.1), doc/src/sgml/release.sgml (1.90.2.1), |
| 108 | +doc/src/sgml/version.sgml (1.3.2.1), src/include/config.h.win32 |
| 109 | +(1.6.2.1), src/interfaces/libpq/libpq.rc (1.7.2.1): Stamp for |
| 110 | +7.1.2.Ready when you are... |
| 111 | + |
| 112 | +2001-05-09 12:28 petere |
| 113 | + |
| 114 | +* src/backend/utils/adt/ri_triggers.c (1.23.2.1): Fix remaining RI |
| 115 | +permission problems (cascaded update/delete, restrict, set |
| 116 | +null/default). |
| 117 | + |
| 118 | +2001-05-08 15:48 tgl |
| 119 | + |
| 120 | +* src/backend/executor/: nodeAppend.c (1.40.2.1), |
| 121 | +nodeSubqueryscan.c (1.6.2.1): Append and SubqueryScan nodes were |
| 122 | +not passing changed-parameter signals down to their children, |
| 123 | +leading to misbehavior if they had any children that paid attention |
| 124 | +to chgParam (most plan node types don't). Append's bug has been |
| 125 | +there a long time, but nobody had noticed because it used to be |
| 126 | +difficult to create a query where an Append would be used below the |
| 127 | +top level of a plan; so there were never any parameters getting |
| 128 | +passed down. SubqueryScan is new in 7.1 ... and I'd modeled its |
| 129 | +behavior on Append :-( |
| 130 | + |
| 131 | +2001-05-07 21:02 tgl |
| 132 | + |
| 133 | +* src/pl/plpgsql/src/pl_exec.c (1.41.2.1): Un-break exec_move_row() |
| 134 | +for case that a NULL tuple and tupdesc are passed, which occurs |
| 135 | +when no rows are retrieved by a SELECT. Mea maxima culpa ... I |
| 136 | +should have caught this. |
| 137 | + |
| 138 | +2001-05-06 17:44 petere |
| 139 | + |
| 140 | +* src/bin/psql/command.c (1.49.2.1): Fix collateral damage from |
| 141 | +previous (rev 1.49) patch. |