mirror of
https://git.postgresql.org/git/postgresql.git
synced 2024-08-21 16:43:24 +02:00
=============== destroying old regression database... ================= =============== creating new regression database... ================= =============== running regression queries... ================= create_function_1 .. ok create_type .. ok create_table .. ok create_function_2 .. ok |
||
---|---|---|
.. | ||
data | ||
expected | ||
input | ||
output | ||
sql | ||
GNUmakefile | ||
Makefile | ||
README | ||
regress.c | ||
regress.sh |
Introduction The PostgreSQL regression tests are a comprehensive set of tests for the SQL implementation embedded in PostgreSQL developed by Jolly Chen and Andrew Yu. It tests standard SQL operations as well as the extensibility capabilities of PostgreSQL. These tests have recently been revised by Marc Fournier and others to become current for PostgreSQL v6.1. The tests are now packaged as functional units and should be easier to run and easier to interpret. Some properly installed and fully functional PostgreSQL installations can fail these regression tests due to artifacts of the genetic optimizer. See the v6.1-specific release notes in this document for further details. Preparation The regression test is invoked by the 'make' command which compiles a 'c' program with PostgreSQL extension functions into a shared library in the current directory. Localized shell scripts are also created in the current directory. The output file templates are massaged into the ./expected/*.out files. The localization replaces macros in the source files with absolute pathnames and user names. The postmaster should be invoked with the system time zone set for Berkeley, California. On some systems, this can be accomplished by setting the TZ environment variable before starting the postmaster (for csh/bash; use set/export for some other shells): setenv TZ PST8PDT7,M04.01.0,M10.05.03 /usr/local/pgsql/bin/postmaster -s Directory Layout input/ .... .source files that are converted using 'make all' into some of the .sql files in the 'sql' subdirectory output/ ... .source files that are converted using 'make all' into .out files in the 'expected' subdirectory sql/ ...... .sql files used to perform the regression tests expected/ . .out files that represent what we *expect* the results to look like results/ .. .out files that represent what the results *actually* look like. Also used as temporary storage for table copy testing. Running the regression test If you have prevously invoked the regression test, clean up the working directory with: make clean The regression test is invoked with the command: make all runtest Normally, the regression test should be run as the pg_superuser as the 'src/test/regress' directory and sub-directories are owned by the pg_superuser. If you run the regression test as another user the 'src/test/regress' directory tree should be writeable to that user. Comparing expected/actual output The results are in the files in the ./results directory. These results can be compared with results in the ./expected directory using 'diff'. The files might not compare exactly. The following paragraphs attempt to explain the differences. OID differences There are several places where PostgreSQL OID (object identifiers) appear in 'regress.out'. OID's are unique 32-bit integers which are generated by the PostgreSQL backend whenever a table row is inserted or updated. If you run the regression test on a non-virgin database or run it multiple times, the OID's reported will have different values. The following SQL statements in 'misc.out' have shown this behavior: QUERY: SELECT user_relns() AS user_relns ORDER BY user_relns; The 'a,523676' row is composed from an OID. TIME differences Some of the tests involving date/time functions use the implicit time zone in effect at the time the regression test is run. In other tests the timezone to be inserted into the regression data base is explicitly specified. The 'expected.input' file was prepared in the 'US/Pacific' timezone so there may be differences where the 'expected.out' file has PST/PDT times and the 'regress.out' file has your local timezone. FLOATING POINT differences Some of the tests involve computing 64-bit (FLOAT8) number from table columns. Differences in results involving mathematical functions of FLOAT8 columns have been observed. These differences occur where different operating systems are used on the same platform ie: BSDI and SOLARIS on Intel/86, and where the same operating system is used used on different platforms, ie: SOLARIS on SPARC and Intel/86. Human eyeball comparison is needed to determine the real significance of these differences which are usually 10 places to the right of the decimal point. POLYGON differences Several of the tests involve operations on geographic date about the Oakland/Berkley CA street map. The map data is expressed as polygons whose vertices are represented as pairs of FLOAT8 numbers (decimal latitude and longitude). Initially, some tables are created and loaded with geographic data, then some views are created which join two tables using the polygon intersection operator (##), then a select is done on the view. When comparing the results from different platforms, differences occur in the 2nd or 3rd place to the right of the decimal point. The SQL statements where these problems occur are the folowing: QUERY: SELECT * from street; QUERY: SELECT * from iexit; DATE/TIME differences On many supported platforms, you can force PostgreSQL to believe that it is running in the same time zone as Berkeley, California. See details in the section on how to run the regression tests. The Makefile attempts to adjust for timezone differences, but it is not possible to totally eliminate them. People outside North America will probabablly find the Makefile's adjustments are incorrect. Also entries that use the time -infinity display with year 1970 plus/minus the number of hours you are different from GMT. Random differences There is at least one test case in misc.out which is intended to produce random results. This causes misc to fail the regression testing. Typing "diff results/misc.out expected/misc.out" should produce only one or a few lines of differences for this reason, but other floating point differences on dissimilar architectures might cause many more differences. The 'expected' files The ./expected/*.out files were adapted from the original monolithic 'expected.input' file provided by Jolly Chen et al. Newer versions of these files generated on various development machines have been substituted after careful (?) inspection. Many of the development machines are running a Unix OS variant (FreeBSD, Linux, etc) on Ix86 hardware. The original 'expected.input' file was created on a SPARC Solaris 2.4 system using the 'postgres5-1.02a5.tar.gz' source tree. It was compared with a file created on an I386 Solaris 2.4 system and the differences were only in the floating point polygons in the 3rd digit to the right of the decimal point. (see below) The original 'sample.regress.out' file was from the postgres-1.01 release constructed by Jolly Chen and is included here for reference. It may have been created on a DEC ALPHA machine as the 'Makefile.global' in the postgres-1.01 release has PORTNAME=alpha. Current release notes There are no release notes for PostgreSQL v6.0. v6.1beta release notes The regression tests have been adapted and extensively modified for the v6.1 release of PostgreSQL. Three new data types (datetime, timespan, and circle) have been added to the native set of PostgreSQL types. Points, boxes, paths, and polygons have had their output formats improved, but the old-style input formats are accepted by v6.1. The source data files have not been updated to the new formats, but should be for the next release. The polygon output in misc.out has only been spot-checked for correctness relative to the original regression output. To get consistant results from the regression tests, compile the PostgreSQL backend with the genetic optimizer (GEQ) turned off. The genetic algorithms introduce a random behavior in the output ordering which causes the simple "diff" implementation of the tests to fail. To turn off the genetic optimizer, edit the src/include/config.h file and comment-out the line containing "#define GEQ", then do a "make clean install" to regenerate the backend. Existing v6.1 databases are not affected by the choice of optimizer, so there is no need to reload after changing the optimizer. The new genetic optimizer has very nice performance with many-table joins, so you may want to make sure to re-enable it and reinstall the code after you have concluded your regression testing. XXX update this for the production release - tgl 97/04/26 The interpretation of array specifiers (the curly braces around atomic values) appears to have changed sometime after the original regression tests were generated. The current ./expected/*.out files reflect this new interpretation, which may not be correct! XXX update this for the production release - tgl 97/04/26 The float8 regression test fails. This may be due to the parser continuing rather than aborting when given invalid constants for input values. XXX update this for the production release - tgl 97/04/26 Regression tests involving indexed tables fail in at least some environments. This may indicate a problem with the current index code.