postgresql82 fails to build on Mac OS X 10.5 Leopard

Me Too.

MacPorts version: 1.520

Last Sync: Sat 27 Oct 2007 04:13:13 UTC

OS: 10.5 (9A581)

nathan@zen: ~ > gcc -v Using built-in specs. Target: i686-apple-darwin9 Configured with: /var/tmp/gcc/gcc-5465~16/src/configure --disable-checking -enable-werror --prefix=/usr --mandir=/share/man --enable-languages=c,objc,c++,obj-c++ --program-transform-name=/^[cg][.-]*$/s/$/-4.0/ --with-gxx-include-dir=/include/c++/4.0.0 --with-slibdir=/usr/lib --build=i686-apple-darwin9 --with-arch=apple --with-tune=generic --host=i686-apple-darwin9 --target=i686-apple-darwin9 Thread model: posix gcc version 4.0.1 (Apple Inc. build 5465)

GNU Make 3.81

Suspect this may be a fat binary issue due to problems I am having linking libpq with from ruby-postgres adapter.

hmm, I get this when building postgresql82:

--->  Building postgresql82
Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake " returned error 2
Command output:   "_boot_yylval", referenced from:
      _boot_yylval$non_lazy_ptr in SUBSYS.o
  "_debug_query_string", referenced from:
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
  "_Geqo_selection_bias", referenced from:
      _Geqo_selection_bias$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesReal in SUBSYS.o
  "_base_yyleng", referenced from:
      _base_yyleng$non_lazy_ptr in SUBSYS.o
  "_locale_time", referenced from:
      _ConfigureNamesString in SUBSYS.o
  "_base_yylloc", referenced from:
      _base_yylloc$non_lazy_ptr in SUBSYS.o
  "_geqo_threshold", referenced from:
      _geqo_threshold$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesInt in SUBSYS.o
  "_boot_yytext", referenced from:
      _boot_yytext$non_lazy_ptr in SUBSYS.o
  "_base_yylval", referenced from:
      _base_yylval$non_lazy_ptr in SUBSYS.o
  "_Dynamic_library_path", referenced from:
      _ConfigureNamesString in SUBSYS.o
      _Dynamic_library_path$non_lazy_ptr in SUBSYS.o
ld: symbol(s) not found
collect2: ld returned 1 exit status
gnumake[2]: *** [postgres] Error 1
gnumake[1]: *** [all] Error 2
gnumake: *** [all] Error 2

I used this patch: http://leenux.org.uk/dtrace-patches/dtrace-with-postgres-on-osx/ added --enable-dtrace and remove the --with-bonjour configure flags and it built for me

Same thing here:

--->  Building postgresql82
Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake " returned error 2
Command output:   "_boot_yylval", referenced from:
      _boot_yylval$non_lazy_ptr in SUBSYS.o
  "_debug_query_string", referenced from:
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
  "_Geqo_selection_bias", referenced from:
      _Geqo_selection_bias$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesReal in SUBSYS.o
  "_base_yyleng", referenced from:
      _base_yyleng$non_lazy_ptr in SUBSYS.o
  "_locale_time", referenced from:
      _ConfigureNamesString in SUBSYS.o
  "_base_yylloc", referenced from:
      _base_yylloc$non_lazy_ptr in SUBSYS.o
  "_geqo_threshold", referenced from:
      _geqo_threshold$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesInt in SUBSYS.o
  "_boot_yytext", referenced from:
      _boot_yytext$non_lazy_ptr in SUBSYS.o
  "_base_yylval", referenced from:
      _base_yylval$non_lazy_ptr in SUBSYS.o
  "_Dynamic_library_path", referenced from:
      _ConfigureNamesString in SUBSYS.o
      _Dynamic_library_path$non_lazy_ptr in SUBSYS.o
ld: symbol(s) not found
collect2: ld returned 1 exit status
gnumake[2]: *** [postgres] Error 1
gnumake[1]: *** [all] Error 2
gnumake: *** [all] Error 2

Error: Status 1 encountered during processing.

Duplicate of #12757?

I've added Lee Packham's dtrace patch to postgresql82 -- please report back if this works;

Still don't work.

--->  Fetching postgresql82
--->  Verifying checksum(s) for postgresql82
--->  Extracting postgresql82
--->  Applying patches to postgresql82
--->  Configuring postgresql82
--->  Building postgresql82
Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake " returned error 2
Command output:   "_boot_yylval", referenced from:
      _boot_yylval$non_lazy_ptr in SUBSYS.o
  "_debug_query_string", referenced from:
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
  "_Geqo_selection_bias", referenced from:
      _Geqo_selection_bias$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesReal in SUBSYS.o
  "_base_yyleng", referenced from:
      _base_yyleng$non_lazy_ptr in SUBSYS.o
  "_locale_time", referenced from:
      _ConfigureNamesString in SUBSYS.o
  "_base_yylloc", referenced from:
      _base_yylloc$non_lazy_ptr in SUBSYS.o
  "_geqo_threshold", referenced from:
      _geqo_threshold$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesInt in SUBSYS.o
  "_boot_yytext", referenced from:
      _boot_yytext$non_lazy_ptr in SUBSYS.o
  "_base_yylval", referenced from:
      _base_yylval$non_lazy_ptr in SUBSYS.o
  "_Dynamic_library_path", referenced from:
      _ConfigureNamesString in SUBSYS.o
      _Dynamic_library_path$non_lazy_ptr in SUBSYS.o
ld: symbol(s) not found
collect2: ld returned 1 exit status
gnumake[2]: *** [postgres] Error 1
gnumake[1]: *** [all] Error 2
gnumake: *** [all] Error 2

same here. This patch doesn't seem to solve the linker problem.

removing the with-bonjour configure flag seems to make it build, however it does seem a bit unstable (crashes for me when doing a big pg_restore) but that looks to be due to my postgis

Here's the crash report (though I guess this should go upstream really):

Process:         postgres [92013]
Path:            /opt/local/lib/postgresql82/bin/postgres
Identifier:      postgres
Version:         ??? (???)
Code Type:       X86 (Native)
Parent Process:  postgres [90]

Date/Time:       2007-10-30 11:11:32.561 +0100
OS Version:      Mac OS X 10.5 (9A581)
Report Version:  6

Exception Type:  EXC_BAD_ACCESS (SIGBUS)
Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000000001
Crashed Thread:  0

Thread 0 Crashed:
0   libSystem.B.dylib             	0x91d7f831 get_uint32 + 32
1   liblwgeom.so                  	0x013f4ea1 lwpoly_deserialize + 161 (lwpoly.c:128)
2   liblwgeom.so                  	0x014062cd LWGEOM_in + 45 (lwgeom_inout.c:75)
3   postgres                      	0x00216491 OidFunctionCall6 + 241
4   postgres                      	0x00216716 OidFunctionCall9 + 22
5   postgres                      	0x00095167 get_rte_attribute_type + 1127
6   postgres                      	0x000976a8 coerce_type + 280
7   postgres                      	0x00097e4f coerce_type + 2239
8   postgres                      	0x00098f12 markTargetListOrigins + 2
9   postgres                      	0x0006f111 transformCreateStmt + 1745
10  postgres                      	0x000728a6 transformInsertStmt + 982
11  postgres                      	0x0006fe04 transformRuleStmt + 2356
12  postgres                      	0x0007146c transformStmt + 5612
13  postgres                      	0x000720aa transformSetOperationStmt + 842
14  postgres                      	0x0018240c errdetail_params + 140
15  postgres                      	0x00182d5c exec_parse_message + 1148
16  postgres                      	0x001848f8 PostgresMain + 4760
17  postgres                      	0x00156dfd ServerLoop + 2045
18  postgres                      	0x00157f4b PostmasterMain + 1995
19  postgres                      	0x0010d11b main + 539
20  postgres                      	0x00001be6 0x1000 + 3046

Thread 0 crashed with X86 Thread State (32-bit):
  eax: 0x00000005  ebx: 0x013f4e11  ecx: 0x00829219  edx: 0x00000001
  edi: 0x0082921d  esi: 0x00829219  ebp: 0xbfffdf78  esp: 0xbfffdf78
   ss: 0x0000001f  efl: 0x00010297  eip: 0x91d7f831   cs: 0x00000017
   ds: 0x0000001f   es: 0x0000001f   fs: 0x00000000   gs: 0x00000037
  cr2: 0x00000001

Binary Images:
    0x1000 -   0x2c4ffb +postgres ??? (???) /opt/local/lib/postgresql82/bin/postgres
  0x35f000 -   0x394feb +libssl.0.9.8.dylib ??? (???) <ca7b9ab31ab9839add8074c2f0661cf2> /opt/local/lib/libssl.0.9.8.dylib
  0x3a7000 -   0x4a9fe7 +libcrypto.0.9.8.dylib ??? (???) <70e5cf8a75766aa004b4a6389460e0ea> /opt/local/lib/libcrypto.0.9.8.dylib
  0x505000 -   0x515ffd +libz.1.dylib ??? (???) /opt/local/lib/libz.1.dylib
  0x597000 -   0x5b0fe3 +plpgsql.so ??? (???) /opt/local/lib/postgresql82/plpgsql.so
  0x5b7000 -   0x5beff7 +libgeos_c.1.dylib ??? (???) /opt/local/lib/libgeos_c.1.dylib
 0x13ec000 -  0x1438fef +liblwgeom.so ??? (???) <3078d2cac1cc65c5a75f7d279dc608ce> /opt/local/lib/postgresql82/liblwgeom.so
 0x145e000 -  0x148efff +libproj.0.dylib ??? (???) /opt/local/lib/libproj.0.dylib
 0x149b000 -  0x152bfff +libgeos.2.dylib ??? (???) /opt/local/lib/libgeos.2.dylib
0x8fe00000 - 0x8fe2d883  dyld 95.3 (???) <81592e798780564b5d46b988f7ee1a6a> /usr/lib/dyld
0x91d20000 - 0x91e7afe3  libSystem.B.dylib ??? (???) <8ecc83dc0399be3946f7a46e88cf4bbb> /usr/lib/libSystem.B.dylib
0x95f8f000 - 0x95f96fe9  libgcc_s.1.dylib ??? (???) <f53c808e87d1184c0f9df63aef53ce0b> /usr/lib/libgcc_s.1.dylib
0x9605a000 - 0x960b7ffb  libstdc++.6.dylib ??? (???) <04b812dcec670daa8b7d2852ab14be60> /usr/lib/libstdc++.6.dylib
0x961bb000 - 0x961bcfef  libmathCommon.A.dylib ??? (???) /usr/lib/system/libmathCommon.A.dylib
0xffff0000 - 0xffff1780  libSystem.B.dylib ??? (???) /usr/lib/libSystem.B.dylib

As discussed on the page with my patch on the blog - although the patch works, it doesn't work within ports. I can only build outside of the port. I'm not sure what's going on with that.

Sorry - need to clarify what I just wrote :)

I can only build when I follow the instructions on http://leenux.org.uk/dtrace-patches/dtrace-with-postgres-on-osx/

I can't build using the port program.

Also - I haven't had any stability issues with postgres.

no, my stability issues are unrelated (postgis related), sorry for cluttering up this ticket folks

removing the with-bonjour configure flag seems to make it build

This does not work for me. Taking out the with-bonjour configure flag doesn't help at all. Is any one else experiencing this?

Also when I compile the code from scratch it configures, builds and installs just fine.

I'm having the same issue.

I'm also having this issue. There appears to be a dtrace patch at Lee Packham's website which I tried but also didn't work in a way that made me wonder if you guys are trying to integrate it.

/me raises hand as well

Same problems here, would be nice to see fixed :)

Some additional info from a clean 10.5 install (not 10.4->10.5 upgrade)

---> Fetching postgresql82 ---> Verifying checksum(s) for postgresql82 ---> Extracting postgresql82 ---> Applying patches to postgresql82 ---> Configuring postgresql82 ---> Building postgresql82 Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake " returned error 2 Command output: "_boot_yylval", referenced from:

_boot_yylval$non_lazy_ptr in SUBSYS.o

"_debug_query_string", referenced from:

_debug_query_string$non_lazy_ptr in SUBSYS.o _debug_query_string$non_lazy_ptr in SUBSYS.o _debug_query_string$non_lazy_ptr in SUBSYS.o

"_Geqo_selection_bias", referenced from:

_Geqo_selection_bias$non_lazy_ptr in SUBSYS.o _ConfigureNamesReal in SUBSYS.o

"_base_yyleng", referenced from:

_base_yyleng$non_lazy_ptr in SUBSYS.o

"_locale_time", referenced from:

_ConfigureNamesString in SUBSYS.o

"_base_yylloc", referenced from:

_base_yylloc$non_lazy_ptr in SUBSYS.o

"_geqo_threshold", referenced from:

_geqo_threshold$non_lazy_ptr in SUBSYS.o _ConfigureNamesInt in SUBSYS.o

"_boot_yytext", referenced from:

_boot_yytext$non_lazy_ptr in SUBSYS.o

"_base_yylval", referenced from:

_base_yylval$non_lazy_ptr in SUBSYS.o

"_Dynamic_library_path", referenced from:

_ConfigureNamesString in SUBSYS.o _Dynamic_library_path$non_lazy_ptr in SUBSYS.o

ld: symbol(s) not found collect2: ld returned 1 exit status gnumake[2]: *** [postgres] Error 1 gnumake[1]: *** [all] Error 2 gnumake: *** [all] Error 2

Error: The following dependencies failed to build: postgresql82 Error: Status 1 encountered during processing.

I spent a little bit of time trying to work around this problem, and discovered that doing a manual 'build' helped. Tried this a couple of times:

sudo port clean postgresql82
sudo port configure postgresql82
cd /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/ work/postgresql-8.2.5
make
cd
sudo port install postgresql82

I did edit out the --with-bonjour, (sudo port edit postgresql82) but that probably didn't help or hinder.

To follow up grady.vincent's comments, I followed the same steps without removing --with-bonjour. I had no problem compiling when I manually cd'd to the work directory and invoked make.

In an attempt to compare the differences, I captured the output of sudo port -d build postgresql82 and compared it to the results of running gnumake in the working directory. There were literally no textual differences in the two log files.

Thinking that it must be a difference in environment, I ran the compile manually in the configured work directory as root using sudo su -. It still compiled fine, and the log files still have literally no text differences.

So, if it isn't a difference in the configure stage, and it isn't different in the environment, what else is different between the port build and executing gnumake by hand. Is there something that port -d build postgresql82 does not report?

Is there something specific about the shell used to execute (taken from the port -d build postgresql82 output):

cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake

That makes it obviously different from:

sudo su -
cd /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5
gnumake

This fixed worked for me. This is interesting though. If I did sudo port install postgresql82 and waited for the error, then went into the directory and typed make as described above, I got the usual error. But when I followed the above list of instructions (sudo port configure postgresql82; ... ; make ; sudo port install postgresql82) it worked. So something is going wrong between the port configure step and the build step, would be my uneducated guess.

I followed the same steps as persson@mac.com and it worked for me as well. I look forward to this working without a manual hack though.

The differnece between a manual gnumake and what port(1) does is, that port(1) not only execs gnumake but also execs gnumake in various subdirectories: adminpack, dblink, fuzzystrmatch, lo, pg_buffercache, pg,_trgm, tsearch2, vacuumlo and xml2 which are add-ons of postgresql that are not part of the default installation.

If someone with a 10.5 installation would mind checking out which contrib is causing this havoc, we can remove it from the installation for 10.5; to do so, just remove items from the line

set contribs "adminpack dblink fuzzystrmatch lo pg_buffercache pg_trgm tsearch2 vacuumlo xml2"

and test if postgresql builds via port(1) then

Replying to mww@macports.org:

If someone with a 10.5 installation would mind checking out which contrib is causing this havoc, we can remove it from the installation for 10.5; to do so, just remove items from the line set contribs "adminpack dblink fuzzystrmatch lo pg_buffercache pg_trgm tsearch2 vacuumlo xml2" and test if postgresql builds via port(1)

I've just tested this and commenting out contribs and this has no effect. The build still fails in the same way.

I can confirm that the steps doc'd by grady.vincent@gmail.com on the 9th worked for me (using the default ports file, ie. I didn't remove bonjour or any of the contrib modules).

Replying to james.croft@lumison.net:

Replying to mww@macports.org:

...

I've just tested this and commenting out contribs and this has no effect. The build still fails in the same way. I can confirm that the steps doc'd by grady.vincent@gmail.com on the 9th worked for me (using the default ports file, ie. I didn't remove bonjour or any of the contrib modules).

would someone please test if adding

system "cd ${worksrcpath} && ${build.cmd}"

as first line after the post-build { statement solves this?

Replying to mww@macports.org: ...

would someone please test if adding system "cd ${worksrcpath} && ${build.cmd}" as first line after the post-build { statement solves this?

no, that wont work -- what I meant was adding:

build.cmd gnumake || gnumake

which would retry to call make IF it fails (once).

now i get:

Error: Target org.macports.build returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5" && gnumake " returned error 2
Command output:   "_boot_yylval", referenced from:
      _boot_yylval$non_lazy_ptr in SUBSYS.o
  "_debug_query_string", referenced from:
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
      _debug_query_string$non_lazy_ptr in SUBSYS.o
  "_Geqo_selection_bias", referenced from:
      _Geqo_selection_bias$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesReal in SUBSYS.o
  "_base_yyleng", referenced from:
      _base_yyleng$non_lazy_ptr in SUBSYS.o
  "_locale_time", referenced from:
      _ConfigureNamesString in SUBSYS.o
  "_base_yylloc", referenced from:
      _base_yylloc$non_lazy_ptr in SUBSYS.o
  "_geqo_threshold", referenced from:
      _geqo_threshold$non_lazy_ptr in SUBSYS.o
      _ConfigureNamesInt in SUBSYS.o
  "_boot_yytext", referenced from:
      _boot_yytext$non_lazy_ptr in SUBSYS.o
  "_base_yylval", referenced from:
      _base_yylval$non_lazy_ptr in SUBSYS.o
  "_Dynamic_library_path", referenced from:
      _ConfigureNamesString in SUBSYS.o
      _Dynamic_library_path$non_lazy_ptr in SUBSYS.o
ld: symbol(s) not found
collect2: ld returned 1 exit status
gnumake[2]: *** [postgres] Error 1
gnumake[1]: *** [all] Error 2
gnumake: *** [all] Error 2

Two days ago, I installed Leopard from scratch and get the exact same errors that you guys do. Please, if you need me to try stuff, just write me an e-mail, so we can get this bug fixed as soon as possible.

the manual build given above by grady vincent also works for me (Leopard/Intel)

I have seen several other ports containing the message " If for some reason this fails, cd to /......blah-blah............. and type make. sorry dont remember where.

this patch to the port file works for me.

--- Portfile.orig	2007-10-30 11:49:08.000000000 +1100
+++ Portfile	2007-11-17 00:42:45.000000000 +1100
@@ -50,6 +50,7 @@
 configure.cppflags-append	"-I${worksrcpath}/src/interfaces/libpq"
 
 build.type		gnu
+build.cmd		"unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}"
 build.target
 
 test.run		yes
@@ -63,7 +64,7 @@
 
 post-build {
 	foreach contrib ${contribs} {
-		system "cd ${worksrcpath}/contrib/${contrib}/ && ${build.cmd}"
+		system "cd ${worksrcpath}/contrib/${contrib}/ && unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && make"
 	}
 }
 
@@ -71,7 +72,7 @@
 	foreach contrib ${contribs} {
 		system "echo contrib: \"${contrib}\""
 		system "cd ${worksrcpath}/contrib/${contrib}/ && \
-			${build.cmd} install ${destroot.destdir} && touch DONE"
+			make install ${destroot.destdir} && touch DONE"
 	}
 	system "cd ${destroot}${prefix}/bin && ln -sf ${libdir}/bin/psql psql82"
 	file delete -force ${destroot}${prefix}/share/doc/${name} \

i can confirm that your patch above works. thanks a lot.

Replying to jodok@lovelysystems.com:

i can confirm that your patch above works. thanks a lot.

Everybody: I just spoke to Jodok, who was so nice as to send me his Portfile (thanks, Jodok!). So this is how you can fix the installation of PostgreSQL 8.2.5 on your clean Leopard install:

1) Download http://www.twothree.dk/macports/postgresql825-fix/Portfile

2) Copy the Portfile you've just downloaded like this:

sudo cp Portfile /opt/local/var/macports/sources/rsync.macports.org/release/ports/databases/postgresql82/

3) You must overwrite the old Portfile with this new one. Now, try to install PostgreSQL 8.2.5 again using MacPorts - it should work.

jodok's patch worked for me as well, thx jodo and kaspert,

Warren Freitag indicated to me in email that the fix in this ticket worked for him too. Are we waiting on something or can the fix be committed?

I'm confused by that path, unless I'm reading it wrong.

It adds/modifies build.cmd to be "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}"

Then it changes two commands to *not* use build.cmd and instead in one instead hard codes "make" ( only once using unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP ) .

It seems like this could be cleaned up to use the variable and that the "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP" is the important part.

Am I missing something?

Replying to grady.vincent@gmail.com:

this patch to the port file works for me.

 +build.cmd		"unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}"
 +		system "cd ${worksrcpath}/contrib/${contrib}/ && unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && make"

Whats the value of LD_PREBIND and LD_PREBIND_ALLOW_OVERLAP on your system? I have no 10.5 installation but I'm surprised those are set by default.

O.k. - this then seems to be a bug of port(1) itself #13436 -- I commited the workaround for the time being;

as I have no way of checking if it builds now, please re-open if the problem persists;

Should that be "set build.cmd"? postgresql82 does not compile for me otherwise.

-build.cmd "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}" +set build.cmd "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}"

Replying to mww@macports.org:

as I have no way of checking if it builds now, please re-open if the problem persists;

With this fix, i get the following error:

Error: Target org.macports.build returned: shell command "cd /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_databases_postgresql82/work/postgresql-8.2.5/contrib/adminpack/ && {unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && gnumake}" returned error 127
Command output: sh: {unset: command not found

Error: Status 1 encountered during processing.

Replying to marcuscalhounlopez@mac.com:

Should that be "set build.cmd"? postgresql82 does not compile for me otherwise. -build.cmd "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}" +set build.cmd "unset LD_PREBIND LD_PREBIND_ALLOW_OVERLAP && ${build.cmd}"

I believe you are correct. I got the same error:

Command output: sh: {unset: command not found

Error: Status 1 encountered during processing.

until I added the 'set' before build.cmd. Now I get it to build correctly.

o.k. - fixed in #31614

The same problem is also found in package postgresql81. The solution is similar, please apply the patch to version 8.1, too.

Replying to mww@macports.org:

o.k. - fixed in #31614

A trivial note: I think you mean [31614], as there is no ticket 31614. :)

fixed in 8.1, too