Обсуждение: mingw 8.1.3 build error: conftest.c: Permission denied

While running the following postgresql-base-8.1.3 configure cmdline in mingw:

./configure --enable-thread-safety --prefix=/cgrid/local/postgresql
--without-zlib

I get the following error(s):

[...]
checking for waitpid... no
checking for wcstombs... yes
checking whether fdatasync is declared... no
checking for struct sockaddr_in6... ./configure: line 13792: conftest.c:
Permission denied
./configure: line 13795: conftest.c: Permission denied
./configure: line 13796: conftest.c: Permission denied
sed: can't read conftest.c: Permission denied
rm: cannot remove `conftest.c': Permission denied
no
checking for PS_STRINGS... ./configure: line 13861: conftest.c: Permission
denied
./configure: line 13864: conftest.c: Permission denied
./configure: line 13865: conftest.c: Permission denied
sed: can't read conftest.c: Permission denied
rm: cannot remove `conftest.c': Permission denied
no
checking whether snprintf is declared... ./configure: line 14177:
conftest.c: Permission denied
./configure: line 14180: conftest.c: Permission denied
./configure: line 14181: conftest.c: Permission denied
sed: can't read conftest.c: Permission denied
rm: cannot remove `conftest.c': Permission denied
no
[...]

And more like them.

Any ideas?  Should this be cause for concern?  The script just continues; I
killed it before proceeding.

Could the fact that /cgrid/local/postgresql does not exist have something
to do with this?  Maybe the --without-zlib?

More details:

$ uname -a
MINGW32_NT-5.1 MATTS-LAPTOP 1.0.11(0.46/3/2) 2004-04-30 18:55 i686 unknown

mengland@MATTS-LAPTOP
/e/data.online/data.cleversafe/svnwork/swdev-external-src/postgresql/dist/mingw/postgresql-8.1.3
$ g++ --version
g++.exe (GCC) 3.4.4 (mingw special)
Copyright (C) 2004 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.


-Matt

At 2/27/2006 11:03 AM, Matt England wrote:
>checking for struct sockaddr_in6... ./configure: line 13792: conftest.c:
>Permission denied
>./configure: line 13795: conftest.c: Permission denied
>./configure: line 13796: conftest.c: Permission denied
>sed: can't read conftest.c: Permission denied
>rm: cannot remove `conftest.c': Permission denied
>no

I should also mentioned that I did the following for my dirs in my build
directory (I'm mengland):

chown -R mengland *
chmod -R u+w *
find . -type d -exec chmod u+rwx \;

...but I haven't yet run the configure with my prefix
(/cgrid/local/postresql) created.

-Matt

Matt England <mengland@mengland.net> writes:
> While running the following postgresql-base-8.1.3 configure cmdline in mingw:
> ./configure --enable-thread-safety --prefix=/cgrid/local/postgresql
> --without-zlib

> I get the following error(s):

> [...]
> checking for waitpid... no
> checking for wcstombs... yes
> checking whether fdatasync is declared... no
> checking for struct sockaddr_in6... ./configure: line 13792: conftest.c:
> Permission denied
> ./configure: line 13795: conftest.c: Permission denied

Interesting ... that test isn't doing anything much that the preceding
tests didn't do.  Maybe some brain-dead "antivirus" code kicking in
to deny file writes?

            regards, tom lane

At 2/27/2006 11:10 AM, Tom Lane wrote:
>Interesting ... that test isn't doing anything much that the preceding
>tests didn't do.

Can I just disregard these errors then?  The ./configure script appears to
proceed ok independent of the error.

>   Maybe some brain-dead "antivirus" code kicking in
>to deny file writes?

Well, I have turned off my the anti-virus file checker for my software that
does such things (PC-Cillin), for whatever that's worth.

-Matt

Matt England <mengland@mengland.net> writes:
> At 2/27/2006 11:10 AM, Tom Lane wrote:
>> Interesting ... that test isn't doing anything much that the preceding
>> tests didn't do.

> Can I just disregard these errors then?  The ./configure script appears to
> proceed ok independent of the error.

Yeah, but it will treat that test as failed which may be (probably is,
in this case) the wrong answer.  In any case, random file-permission
failures constitute a platform instability that you do not want to just
live with.  You might have trouble getting through the build step without
more of them, to take just one point.

            regards, tom lane