Build report for 6.4 BETA2 on Digital Unix 4.0d

Hi, all.

I've just compiled 6.4 BETA2 on my DU 4.0d system with DEC C, and have
also run the regression tests.

I had the following problems when building:

1 - The DEC C preprocessor doesn't accept input from stdin. configure
tries to use "cc -E -", but it doesn't work. According to Tom Lane, this
should be fixed in BETA4. I worked around this problem patching
Gen_fmgrtab.sh.in and genbki.sh.in as suggested by Thomas Lockhart in an
earlier message.

2 - I got many warnings complaining that MAP_FILE was redefined. Indeed,
MAP_FILE is defined by many systems in "sys/mman.h", as it is used with
mmap(2) and friends. MAP_FILE in include/libpq/hba.h should be renamed to
something like PG_MAP_FILE.

3 - I had to add the option '-std' to the compiler flags when compiling in
interfaces/libpq. This one is platform-specific.

4 - I had to edit Makefile.tcldefs in pl/tcl. Digital Unix uses 'ld
-shared -expect_unresolved "*"' to generate shared libraries, and so it is
specified in my system's /usr/local/lib/tclConfig.sh. However, somwhere in
the config process the quotes around the asterisk dissappeared, and so the
shell expanded the asterisk. I had to put the quotes back. I didn't have
the time to investigate more.

The following regression tests failed:

- char, varchar

They failed because for some obscure reason 'A' is considered greater than
'a'. I won't be able to investigate on this until tomorrow.

- int2, int4

They failed because of differences in the expected error message, so they
are ok.

- int8

The int8 type is completely brain damaged on my system. Strange, because
Digital Unix is a true 64 bit system. Again, I'll try to look at it
tomorrow.

- float8

It had one error-message-format error (not really an error) and one real
error ("exp() result is out of range" when no error was expected).

- abstime, tinterval, horology

Seems to be a time zone problem, though I ran the test with TZ=PST8PDT. I
haven't had time to investigate deeper.

- select_implicit, select_having, random, rules, plpgsql

I haven't had time to look at these. I'll look at them tomorrow.

Greetings,

Pedro.

-------------------------------------------------------------------
Pedro Jos� Lobo Perea Tel: +34 91 336 78 19
Centro de C�lculo Fax: +34 91 331 92 29
EUIT Telecomunicaci�n - UPM e-mail: pjlobo@euitt.upm.es