[GRASS-dev] Re: [bug #5185] shared library problem solaris

Hi grass-dev,

On Mon, May 14, 2007 at 02:12:03PM +0200, Request Tracker wrote:

this bug's URL: http://intevation.de/rt/webrt?serial_num=5185

Mon, May 14 2007 14:12:03: Request 5185 was acted upon.

Transaction: Mail sent by harri.kiiskinen utu.fi

...

> Harri, could you please test if it now works?

Sorry to say, but it does not yet work; I do have a suspicion, that the
test in aclocal.m4 on line 1137 is the reason: `uname -s`-`uname -r`
(aclocal.m4 line 446) gives 'SunOS-5.10' on the system I'm using, and
the results do indicate, that contrary to what planned, the system is
treated as on of 5.0 to 5.6.

Harri K.

... new ideas are welcome again...

Markus

On Mon, 14 May 2007, Markus Neteler wrote:

Sorry to say, but it does not yet work; I do have a suspicion, that the
test in aclocal.m4 on line 1137 is the reason: `uname -s`-`uname -r`
(aclocal.m4 line 446) gives 'SunOS-5.10' on the system I'm using, and
the results do indicate, that contrary to what planned, the system is
treated as on of 5.0 to 5.6.

Harri K.

... new ideas are welcome again...

Looks like it is treating 5.1.* and 5.10* the same - adding an extra dot
into the matching expression should fix it I would hope:

Index: aclocal.m4

RCS file: /home/grass/grassrepository/grass6/aclocal.m4,v
retrieving revision 1.32
diff -u -r1.32 aclocal.m4
--- aclocal.m4 12 May 2007 15:03:33 -0000 1.32
+++ aclocal.m4 14 May 2007 15:24:14 -0000
@@ -1134,7 +1134,7 @@
             UNSHARED_LIB_SUFFIX='${GRASS_TRIM_DOTS}.a'
             GRASS_LIB_VERSIONS_OK=nodots
             ;;
- SunOS-5.[[0-6]]*)
+ SunOS-5.[[0-6]].*)

             # Note: If _REENTRANT isn't defined, then Solaris
             # won't define thread-safe library routines.

On Mon, May 14, 2007 at 04:29:02PM +0100, Paul Kelly wrote:

On Mon, 14 May 2007, Markus Neteler wrote:

>>Sorry to say, but it does not yet work; I do have a suspicion, that the
>>test in aclocal.m4 on line 1137 is the reason: `uname -s`-`uname -r`
>>(aclocal.m4 line 446) gives 'SunOS-5.10' on the system I'm using, and
>>the results do indicate, that contrary to what planned, the system is
>>treated as on of 5.0 to 5.6.
>>
>>Harri K.
>
>... new ideas are welcome again...

Looks like it is treating 5.1.* and 5.10* the same - adding an extra dot
into the matching expression should fix it I would hope:

Index: aclocal.m4

RCS file: /home/grass/grassrepository/grass6/aclocal.m4,v
retrieving revision 1.32
diff -u -r1.32 aclocal.m4
--- aclocal.m4 12 May 2007 15:03:33 -0000 1.32
+++ aclocal.m4 14 May 2007 15:24:14 -0000
@@ -1134,7 +1134,7 @@
            UNSHARED_LIB_SUFFIX='${GRASS_TRIM_DOTS}.a'
            GRASS_LIB_VERSIONS_OK=nodots
            ;;
- SunOS-5.[[0-6]]*)
+ SunOS-5.[[0-6]].*)

            # Note: If _REENTRANT isn't defined, then Solaris
            # won't define thread-safe library routines.

Thanks, Paul. In CVS now.

Harri, please try again...
Markus

Paul Kelly wrote:

>> Sorry to say, but it does not yet work; I do have a suspicion, that the
>> test in aclocal.m4 on line 1137 is the reason: `uname -s`-`uname -r`
>> (aclocal.m4 line 446) gives 'SunOS-5.10' on the system I'm using, and
>> the results do indicate, that contrary to what planned, the system is
>> treated as on of 5.0 to 5.6.
>>
>> Harri K.
>
> ... new ideas are welcome again...

Looks like it is treating 5.1.* and 5.10* the same - adding an extra dot
into the matching expression should fix it I would hope:

This might be a good time to think about eliminating the use of
"uname" and using $host like it should.

--
Glynn Clements <glynn@gclements.plus.com>