Re: pgsql: Fix install program detection

Fix install program detection

configure handles INSTALL as a substitution variable specially, and
apparently it gets confused when it's set to empty. Use INSTALL_
instead as a workaround to avoid the issue.

Branch
------
master

Details
-------
http://git.postgresql.org/pg/commitdiff/f7867154129781ee1522344bef50890c01f2b47a

Modified Files
--------------
configure | 5 ++++-
configure.in | 4 +++-
src/Makefile.global.in | 2 +-
3 files changed, 8 insertions(+), 3 deletions(-)

Excerpts from Peter Eisentraut's message of mié jun 27 14:23:11 -0400 2012:
> Fix install program detection
>
> configure handles INSTALL as a substitution variable specially, and
> apparently it gets confused when it's set to empty. Use INSTALL_
> instead as a workaround to avoid the issue.

Member grebe is still failing. Interesting log lines:

/bin/sh ../../config/install-sh -c -d '/opt/rg/data_dba/build-farm/HEAD/pgsql.401504/src/test/regress/./tmp_check/install/opt/rg/data_dba/build-farm/HEAD/inst/bin' '/opt/rg/data_dba/build-farm/HEAD/pgsql.401504/src/test/regress/./tmp_check/install/opt/rg/data_dba/build-farm/HEAD/inst/share/postgresql'
/bin/sh ../../config/install-sh -c -d '/opt/rg/data_dba/build-farm/HEAD/pgsql.401504/src/test/regress/./tmp_check/install/opt/rg/data_dba/build-farm/HEAD/inst/lib/postgresql'
postgres '/opt/rg/data_dba/build-farm/HEAD/pgsql.401504/src/test/regress/./tmp_check/install/opt/rg/data_dba/build-farm/HEAD/inst/bin/postgres'
gmake[3]: postgres: Command not found
gmake[3]: *** [install-bin] Error 127

Evidently the logic to set INSTALL_PROGRAM is wrong on that platform ...

--
Álvaro Herrera <alvherre(at)commandprompt(dot)com>
The PostgreSQL Company - Command Prompt, Inc.
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> Excerpts from Peter Eisentraut's message of mi jun 27 14:23:11 -0400 2012:
>> Fix install program detection

> Member grebe is still failing.

protosciurus and castoroides (Solaris 10, I think same machine with
different compilers) are not happy either. Also, the first failure
on those looks like:

gmake[4]: Leaving directory `/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/backend'
/bin/sh ../../config/install-sh -c -d '/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/test/regress/./tmp_check/install/export/home/dpage/pgbuildfarm/castoroides/HEAD/inst/lib'
m 644 libpgport.a '/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/test/regress/./tmp_check/install/export/home/dpage/pgbuildfarm/castoroides/HEAD/inst/lib/libpgport.a'
gmake[3]: m: Command not found
gmake[3]: [install] Error 127 (ignored)

which seems bad for another reason: why are we ignoring a command error there?

regards, tom lane

On Thu, Jun 28, 2012 at 12:28 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
>> Excerpts from Peter Eisentraut's message of mié jun 27 14:23:11 -0400 2012:
>>> Fix install program detection
>
>> Member grebe is still failing.
>
> protosciurus and castoroides (Solaris 10, I think same machine with
> different compilers) are not happy either.  Also, the first failure
> on those looks like:
>
> gmake[4]: Leaving directory `/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/backend'
> /bin/sh ../../config/install-sh -c -d '/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/test/regress/./tmp_check/install/export/home/dpage/pgbuildfarm/castoroides/HEAD/inst/lib'
> m 644  libpgport.a '/export/home/dpage/pgbuildfarm/castoroides/HEAD/pgsql.17031/src/test/regress/./tmp_check/install/export/home/dpage/pgbuildfarm/castoroides/HEAD/inst/lib/libpgport.a'
> gmake[3]: m: Command not found
> gmake[3]: [install] Error 127 (ignored)
>
> which seems bad for another reason: why are we ignoring a command error there?

I'm thinking that $INSTALL or whatever-it-is is getting set to an
empty string. The command was probably intended to be something like
install -m 644 libpgport.a...

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> I'm thinking that $INSTALL or whatever-it-is is getting set to an
> empty string. The command was probably intended to be something like
> install -m 644 libpgport.a...

So it appears, but why is the failure platform-specific? INSTALL is
also getting set to empty on my old HPUX box, but the install goes
through OK there.

regards, tom lane