22Introduction
33
44 The PostgreSQL regression tests are a comprehensive set of tests for the
5- SQL implementation embeded in PostgreSQL developed by Jolly Chen and
6- Andrew Yu. It tests standard SQL operations as well as the extensability
5+ SQL implementation embedded in PostgreSQL developed by Jolly Chen and
6+ Andrew Yu. It tests standard SQL operations as well as the extensibility
77 capabilities of PostgreSQL.
88
9+ These tests have recently been revised by Marc Fournier and others to
10+ become current for PostgreSQL v6.1. The tests are now packaged as
11+ functional units and should be easier to run and easier to interpret.
12+
13+ Some properly installed and fully functional PostgreSQL installations
14+ can fail these regression tests due to artifacts of the genetic optimizer.
15+ See the v6.1-specific release notes in this document for further details.
16+
917Preparation
1018
11- The regression test is invoked thru by the 'make' command which compiles
19+ The regression test is invoked by the 'make' command which compiles
1220 a 'c' program with PostgreSQL extension functions into a shared library
1321 in the current directory. Localized shell scripts are also created in
14- the current directory. The 'expected.input' file is massaged into the
15- ' expected.out' file . The localization replaces macros in the source
22+ the current directory. The output file templates are massaged into the
23+ ./ expected/* .out files . The localization replaces macros in the source
1624 files with absolute pathnames and user names.
1725
26+ The postmaster should be invoked with the system time zone set for
27+ Berkeley, California. On some systems, this can be accomplished by
28+ setting the TZ environment variable before starting the postmaster
29+ (for csh/bash; use set/export for some other shells):
30+
31+ setenv TZ PST8PDT7,M04.01.0,M10.05.03
32+ /usr/local/pgsql/bin/postmaster -s
33+
1834Directory Layout
1935
2036 input/ .... .source files that are converted using 'make all' into
21- .sql files in the 'sql' subdirectory
37+ some of the .sql files in the 'sql' subdirectory
2238
2339 output/ ... .source files that are converted using 'make all' into
2440 .out files in the 'expected' subdirectory
@@ -29,7 +45,7 @@ Directory Layout
2945 look like
3046
3147 results/ .. .out files that represent what the results *actually* look
32- like
48+ like. Also used as temporary storage for table copy testing.
3349
3450Running the regression test
3551
@@ -45,14 +61,14 @@ Running the regression test
4561 Normally, the regression test should be run as the pg_superuser as the
4662 'src/test/regress' directory and sub-directories are owned by the
4763 pg_superuser. If you run the regression test as another user the
48- 'src/test/regress' directory should be writeable to that user.
64+ 'src/test/regress' directory tree should be writeable to that user.
4965
5066Comparing expected/actual output
5167
52- The results are in the file 'regress.out' which can be compared
53- with the 'expected.out' file using 'diff'. The files will NOT
54- compare exactly . The following paragraphs attempt to explain the
55- differences.
68+ The results are in the files in the ./results directory. These
69+ results can be compared with results in the ./expected directory
70+ using 'diff' . The files might not compare exactly. The following
71+ paragraphs attempt to explain the differences.
5672
5773OID differences
5874
@@ -62,7 +78,7 @@ OID differences
6278 If you run the regression test on a non-virgin database or run it multiple
6379 times, the OID's reported will have different values.
6480
65- The following SQL statements in 'regress .out' have shown this behavior:
81+ The following SQL statements in 'misc .out' have shown this behavior:
6682
6783 QUERY: SELECT user_relns() AS user_relns ORDER BY user_relns;
6884
@@ -96,8 +112,8 @@ POLYGON differences
96112
97113 Several of the tests involve operations on geographic date about the
98114 Oakland/Berkley CA street map. The map data is expressed as polygons
99- whose verticies are represened as pairs of FLOAT8 numbers (decimal
100- lattitude and longitude). Initially, some tables are created and
115+ whose vertices are represented as pairs of FLOAT8 numbers (decimal
116+ latitude and longitude). Initially, some tables are created and
101117 loaded with geographic data, then some views are created which join
102118 two tables using the polygon intersection operator (##), then a select
103119 is done on the view.
@@ -111,24 +127,84 @@ POLYGON differences
111127
112128DATE/TIME differences
113129
130+ On many supported platforms, you can force PostgreSQL to believe that it
131+ is running in the same time zone as Berkeley, California. See details in
132+ the section on how to run the regression tests.
133+
114134 The Makefile attempts to adjust for timezone differences, but it is
115- totally possible to eliminate them. People outside North America
135+ not possible to totally eliminate them. People outside North America
116136 will probabablly find the Makefile's adjustments are incorrect. Also
117137 entries that use the time -infinity display with year 1970 plus/minus the
118138 number of hours you are different from GMT.
119139
120- --------[ old stuff that needs to be rewritten ]-----
140+ Random differences
141+
142+ There is at least one test case in misc.out which is intended to produce
143+ random results. This causes misc to fail the regression testing.
144+ Typing "diff results/misc.out expected/misc.out" should produce only
145+ one or a few lines of differences for this reason, but other floating
146+ point differences on dissimilar architectures might cause many more
147+ differences.
121148
122- The 'expected.input' file and the 'sample.regress.out' file
149+ The 'expected' files
123150
124- The 'expected.input' file was created on a SPARC Solaris 2.4 system
125- using the 'postgres5-1.02a5.tar.gz' source tree. It has been compared
151+ The ./expected/*.out files were adapted from the original monolithic
152+ 'expected.input' file provided by Jolly Chen et al. Newer versions of these
153+ files generated on various development machines have been substituted after
154+ careful (?) inspection. Many of the development machines are running a
155+ Unix OS variant (FreeBSD, Linux, etc) on Ix86 hardware.
156+
157+ The original 'expected.input' file was created on a SPARC Solaris 2.4
158+ system using the 'postgres5-1.02a5.tar.gz' source tree. It was compared
126159 with a file created on an I386 Solaris 2.4 system and the differences
127- are only in the floating point polygons in the 3rd digit to the right
160+ were only in the floating point polygons in the 3rd digit to the right
128161 of the decimal point. (see below)
129162
130- The 'sample.regress.out' file is from the postgres-1.01 release
163+ The original 'sample.regress.out' file was from the postgres-1.01 release
131164 constructed by Jolly Chen and is included here for reference. It may
132165 have been created on a DEC ALPHA machine as the 'Makefile.global'
133166 in the postgres-1.01 release has PORTNAME=alpha.
134167
168+ Current release notes
169+
170+ There are no release notes for PostgreSQL v6.0.
171+
172+ v6.1beta release notes
173+
174+ The regression tests have been adapted and extensively modified for the
175+ v6.1 release of PostgreSQL.
176+
177+ Three new data types (datetime, timespan, and circle) have been added to
178+ the native set of PostgreSQL types. Points, boxes, paths, and polygons
179+ have had their output formats improved, but the old-style input formats
180+ are accepted by v6.1. The source data files have not been updated to the
181+ new formats, but should be for the next release. The polygon output in
182+ misc.out has only been spot-checked for correctness relative to the
183+ original regression output.
184+
185+ To get consistant results from the regression tests, compile the PostgreSQL
186+ backend with the genetic optimizer (GEQ) turned off. The genetic algorithms
187+ introduce a random behavior in the output ordering which causes the
188+ simple "diff" implementation of the tests to fail. To turn off the genetic
189+ optimizer, edit the src/include/config.h file and comment-out the line
190+ containing "#define GEQ", then do a "make clean install" to regenerate
191+ the backend. Existing v6.1 databases are not affected by the choice of
192+ optimizer, so there is no need to reload after changing the optimizer.
193+ The new genetic optimizer has very nice performance with many-table joins,
194+ so you may want to make sure to re-enable it and reinstall the code after
195+ you have concluded your regression testing.
196+
197+ XXX update this for the production release - tgl 97/04/26
198+ The interpretation of array specifiers (the curly braces around atomic
199+ values) appears to have changed sometime after the original regression
200+ tests were generated. The current ./expected/*.out files reflect this
201+ new interpretation, which may not be correct!
202+
203+ XXX update this for the production release - tgl 97/04/26
204+ The float8 regression test fails. This may be due to the parser continuing
205+ rather than aborting when given invalid constants for input values.
206+
207+ XXX update this for the production release - tgl 97/04/26
208+ Regression tests involving indexed tables fail in at least some environments.
209+ This may indicate a problem with the current index code.
210+
0 commit comments