I have run into the same sequence of problems as Juhan Ernits,
and I still do not have a solution.
I just tried a few commands that might shed some light on the problem of
bash-2.04$ nm /lib/libc.a | grep errno
00000000 ? __imp__h_errno
00000000 T ___errno
00000000 ? __imp____errno
bash-2.04$ nm /lib/libc.a | grep sys_nerr
00000000 ? __imp___sys_nerr
We see that the C library does define errno, but not sys_nerr.
Searching the other libraries I found that sys_nerr and sys_errlist are both
in the libraries /usr/lib/libmsvcrt20.a and /usr/lib/libmsvcrt40.a
I do not know what these libraries are for, or why sys_errlist is not in
does anyone have a hint?
(Oh, could be that msvcrt stands for Microsoft Visual C Runtime?)
If libc does not have sys_errlist, where does strerror() get its data from?
Should we rather rely on strerror to check the validity of the argument,
and output the alternative string only if strerror returns a null pointer
or a pointer to an empty string?
Juhan Ernits <Juhan(at)suhkur(dot)cc(dot)ioc(dot)ee> wrote:
>System is NT4 SP6, cygwin 1.1, gcc 2.95.2, PG7RC2.
>Having fixed the dllmain problem, I got the following one:
>The step that assembles postgres.exe fails with the following messages:
>utils/SUBSYS.o(text+0x42de7):elog.c: undefined reference to '_sys_nerr'
>utils/SUBSYS.o(text+0x437b3):exc.c: undefined reference to '_sys_nerr'
>collect2: ld returned 1 exit status
>make: *** [postgres] Error 1
>How should one cope with that?
pgsql-ports by date
|Next:||From: Ryan Kirkpatrick||Date: 2000-06-18 22:43:03|
|Subject: Re: Call for port testing on fmgr changes -- Results!|
|Previous:||From: Doug Edmunds||Date: 2000-06-16 17:21:04|
|Subject: Regression test results not the same |