Hi:

I am having a few problems compiling beta 10.

My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
to get betas up through beta 9 to work.

1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
FP_PLUS_INF. I just copy them from the system include file.

2. I have to manually set the database type (aix-gcc) on configuration.

(The two previous problems are old, and are as second nature as setting
the execute bit on the config scripts.)

And now for the new one:

The main linking step seems to be looking for a libgcc.a! I haven't changed
my system since I compiled bata 9 (yesterday), and I don't know why it isn't
finding it. It exists in the proper directory:
/usr/local/lib/gcc-lib/powerpc-ibm-aix4.1.5.0/2.7.2.2

It never seemed to have trouble finding this directory before!

The actual error is:

Making postgres.imp
../backend/port/aix/mkldexport.sh postgres /usr/local/pgsql/bin > postgres.imp
gcc -bE:../backend/postgres.imp -o postgres access/SUBSYS.o bootstrap/SUBSYS.o c
atalog/SUBSYS.o commands/SUBSYS.o executor/SUBSYS.o lib/SUBSYS.o libpq/SUBSYS.o
main/SUBSYS.o nodes/SUBSYS.o optimizer/SUBSYS.o parser/SUBSYS.o port/SUBSYS.o po
stmaster/SUBSYS.o regex/SUBSYS.o rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o
utils/SUBSYS.o ../utils/version.o -lPW -lld -lm -lbsd -ltermcap -lcurses
ld: 0706-005 Cannot find or open file: libgcc.a
ld:open(): No such file or directory
ld: 0706-005 Cannot find or open file: libgcc.a
ld:open(): No such file or directory

I will try configuring the system with the directoy explicitely listed
and see what happens.
- --
D. Jay Newman ! For the pleasure and the profit it derives
jay@sprucegrove.com ! I arrange things, like furniture, and
http://www.sprucegrove.com/~jay/ ! daffodils, and ...lives. -- Hello Dolly

------------------------------

Search Discussions

  • Bruce Momjian at Sep 23, 1997 at 9:08 pm
    Please check the file float.c, and see the AIX define just above the
    HAVE_INF. Tell us the name of the include file, and I will put it in
    the file at that point.
    Hi:

    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.

    2. I have to manually set the database type (aix-gcc) on configuration.
    This should have been fixed in beta10. Please run config.guess, and
    tell me what is says. No one has sent us an AIX output for this.
    (The two previous problems are old, and are as second nature as setting
    the execute bit on the config scripts.)

    And now for the new one:

    The main linking step seems to be looking for a libgcc.a! I haven't changed
    my system since I compiled bata 9 (yesterday), and I don't know why it isn't
    finding it. It exists in the proper directory:
    /usr/local/lib/gcc-lib/powerpc-ibm-aix4.1.5.0/2.7.2.2

    It never seemed to have trouble finding this directory before!

    The actual error is:

    Making postgres.imp
    ../backend/port/aix/mkldexport.sh postgres /usr/local/pgsql/bin > postgres.imp
    gcc -bE:../backend/postgres.imp -o postgres access/SUBSYS.o bootstrap/SUBSYS.o c
    atalog/SUBSYS.o commands/SUBSYS.o executor/SUBSYS.o lib/SUBSYS.o libpq/SUBSYS.o
    main/SUBSYS.o nodes/SUBSYS.o optimizer/SUBSYS.o parser/SUBSYS.o port/SUBSYS.o po
    stmaster/SUBSYS.o regex/SUBSYS.o rewrite/SUBSYS.o storage/SUBSYS.o tcop/SUBSYS.o
    utils/SUBSYS.o ../utils/version.o -lPW -lld -lm -lbsd -ltermcap -lcurses
    ld: 0706-005 Cannot find or open file: libgcc.a
    ld:open(): No such file or directory
    ld: 0706-005 Cannot find or open file: libgcc.a
    ld:open(): No such file or directory

    I will try configuring the system with the directoy explicitely listed
    and see what happens.
    I don't know what would cause this.

    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Gerhard Reithofer at Sep 24, 1997 at 7:11 am

    On Tue, 23 Sep 1997, Bruce Momjian wrote:

    Please check the file float.c, and see the AIX define just above the
    HAVE_INF. Tell us the name of the include file, and I will put it in
    the file at that point.
    Hi:

    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    ...

    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.

    Cheers,
    Gerhard

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    ------------------------------
  • Darren King at Sep 24, 1997 at 1:30 pm

    Please check the file float.c, and see the AIX define just above the
    HAVE_INF. Tell us the name of the include file, and I will put it in
    the file at that point.
    The name of the include file is <float.h>.
    Hi:

    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.

    2. I have to manually set the database type (aix-gcc) on configuration.
    This should have been fixed in beta10. Please run config.guess, and
    tell me what is says. No one has sent us an AIX output for this.
    The output of config.guess is:
    powerpc-ibm-aix4.1.5.0

    This is probably because it is an RS/6000 series running on a PowerPC chip.
    A nice desktop unix box.
    Correct.

    The two aix boxes here give "powerpc-ibm-aix3.2.5" and "powerpc-ibm-aix4.1.4.0".
    Adding these and yours for 4.1.5 to the .similar file would be easy, but which
    template do you point it towards?

    I want to leave the aix-gcc template and then create one for 325 and 41 from the
    aix-cc template.

    Is there still time to get this in the 6.2 release? I can have it in a couple
    of hours...involves just a little fudging the templates and Makefile.aix.


    Darren darrenk@insightdist.com

    ------------------------------
  • D. Jay Newman at Sep 24, 1997 at 2:09 pm
    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.

    Cheers,
    Gerhard
    - --
    D. Jay Newman ! For the pleasure and the profit it derives
    jay@sprucegrove.com ! I arrange things, like furniture, and
    http://www.sprucegrove.com/~jay/ ! daffodils, and ...lives. -- Hello Dolly

    ------------------------------
  • Bruce Momjian at Sep 24, 1997 at 3:38 pm
    Would the person who reported this let us know if it is true. If so, I
    need to reverse my patch.

    I am reversing my addition of float.h to adt/float.c because I now see
    it is already there.
    On Tue, 23 Sep 1997, Bruce Momjian wrote:

    Please check the file float.c, and see the AIX define just above the
    HAVE_INF. Tell us the name of the include file, and I will put it in
    the file at that point.
    Hi:

    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    ...

    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.

    Cheers,
    Gerhard

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Bruce Momjian at Sep 24, 1997 at 3:48 pm

    Please check the file float.c, and see the AIX define just above the
    HAVE_INF. Tell us the name of the include file, and I will put it in
    the file at that point.
    The name of the include file is <float.h>.
    Hi:

    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.

    2. I have to manually set the database type (aix-gcc) on configuration.
    This should have been fixed in beta10. Please run config.guess, and
    tell me what is says. No one has sent us an AIX output for this.
    The output of config.guess is:
    powerpc-ibm-aix4.1.5.0

    This is probably because it is an RS/6000 series running on a PowerPC chip.
    A nice desktop unix box.
    Correct.

    The two aix boxes here give "powerpc-ibm-aix3.2.5" and "powerpc-ibm-aix4.1.4.0".
    Adding these and yours for 4.1.5 to the .similar file would be easy, but which
    template do you point it towards?

    I want to leave the aix-gcc template and then create one for 325 and 41 from the
    aix-cc template.

    Is there still time to get this in the 6.2 release? I can have it in a couple
    of hours...involves just a little fudging the templates and Makefile.aix.
    I am a little confused by this. I don't think we support the power pc
    chip, do we? We support aix, but that is on another chipset. You
    normally don't need a new template for every OS release, and configure
    will find a match without the version number if an exact one does not
    exist.


    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Bruce Momjian at Sep 24, 1997 at 3:51 pm
    You are the only one who is going to be able to help us. Can we modify
    the aix-ifdef in adt/float.c to only trigger if use use cc and aix, and
    not gcc and aix?
    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.

    Cheers,
    Gerhard
    --
    D. Jay Newman ! For the pleasure and the profit it derives
    jay@sprucegrove.com ! I arrange things, like furniture, and
    http://www.sprucegrove.com/~jay/ ! daffodils, and ...lives. -- Hello Dolly

    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Darren King at Sep 24, 1997 at 4:03 pm

    Bruce wrote:

    You are the only one who is going to be able to help us. Can we modify
    the aix-ifdef in adt/float.c to only trigger if use use cc and aix, and
    not gcc and aix?
    How about if aix and not using gcc? The IBM compiler suite has many variations,
    cc, xlc, xlC, ... eight different I believe. They are all aliased back to cc
    though with different args. I've found that xlc or xlC works best for compiling
    postgres, so keying off of "cc" could be troublesome.


    darrenk

    ------------------------------
  • Bruce Momjian at Sep 24, 1997 at 4:56 pm

    Bruce wrote:

    You are the only one who is going to be able to help us. Can we modify
    the aix-ifdef in adt/float.c to only trigger if use use cc and aix, and
    not gcc and aix?
    How about if aix and not using gcc? The IBM compiler suite has many variations,
    cc, xlc, xlC, ... eight different I believe. They are all aliased back to cc
    though with different args. I've found that xlc or xlC works best for compiling
    postgres, so keying off of "cc" could be troublesome.
    Keying off of gcc or non-gcc is a good idea.

    The issue is if you don't need to run the aix-specific code for gcc, we
    are ok, and can add a non-gcc test to adt/float.c.

    If the code is needed for aix in gcc or non-gcc, you will have to
    include /usr/include/float.h directly, rather than using the include
    search path.

    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Gerhard Reithofer at Sep 24, 1997 at 7:52 pm

    On Wed, 24 Sep 1997, D. Jay Newman wrote:

    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.
    I'm thinking of that problem very long - and I only know the (ugly?) way:
    Add the entries:
    -DFP_PLUS_INF=4 -DFP_MINUS_INF=5
    to the compiler flag switches in the aix-gcc template.

    Bye,
    Gerhard

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    ------------------------------
  • Bruce Momjian at Sep 24, 1997 at 8:19 pm

    On Wed, 24 Sep 1997, D. Jay Newman wrote:

    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    I am having a few problems compiling beta 10.

    My system is a 43p running AIX 4.1.5 and compiling with gcc. I've been able
    to get betas up through beta 9 to work.

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.
    I'm thinking of that problem very long - and I only know the (ugly?) way:
    Add the entries:
    -DFP_PLUS_INF=4 -DFP_MINUS_INF=5
    to the compiler flag switches in the aix-gcc template.
    Or, we can add that into float.h if gcc if using aix, and gcc is being
    defined, or you could try:

    #include </usr/include/float.h>

    in the aix-specific code in float.c.

    Does that work? It over-rides the search for includes.


    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Thomas G. Lockhart at Sep 24, 1997 at 11:52 pm

    1. backend/utils/adt/float.c needs the definitions of FP_MINUS_INF, and
    FP_PLUS_INF. I just copy them from the system include file.
    IMHO the problem is, that GCC installs his own float.h which does not
    include the proper #define's but on the AIX side a float.h exists with the
    #define's.
    Configure finds the AIX one and GCC does not use it.
    I'm thinking of that problem very long - and I only know the (ugly?) way:
    Add the entries:
    -DFP_PLUS_INF=4 -DFP_MINUS_INF=5
    to the compiler flag switches in the aix-gcc template.
    Or, we can add that into float.h if gcc if using aix, and gcc is being
    defined, or you could try:

    #include </usr/include/float.h>

    in the aix-specific code in float.c.
    How about

    #ifndef FP_PLUS_INF
    #define FP_PLUS_INF 4
    #endif
    etc.

    somewhere in the source code, perhaps in aix.h?

    - Tom

    ------------------------------
  • Gerhard Reithofer at Sep 25, 1997 at 8:43 am
    This message is in MIME format. The first part should be readable text,
    while the remaining parts are likely unreadable without MIME-aware tools.
    Send mail to mime@docserver.cac.washington.edu for more info.

    - ---1463809280-1334800097-875177029=:1218
    Content-Type: TEXT/PLAIN; charset=US-ASCII
    On Wed, 24 Sep 1997, Bruce Momjian wrote:

    On Wed, 24 Sep 1997, D. Jay Newman wrote:

    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    ... CUT ...
    I'm thinking of that problem very long - and I only know the (ugly?) way:
    Add the entries:
    -DFP_PLUS_INF=4 -DFP_MINUS_INF=5
    to the compiler flag switches in the aix-gcc template.
    Or, we can add that into float.h if gcc if using aix, and gcc is being
    defined, or you could try:

    #include </usr/include/float.h>

    in the aix-specific code in float.c.

    Does that work? It over-rides the search for includes.
    No very good idea, it produces many double-defined warnings.

    My suggestion (I think, the current implementation does not produce
    correct code with gcc?) is to implement a aixgcc version of isinf - it'll
    be not the only one ;-).

    Example:
    #if defined(aixgcc)
    static int isinf(double x)
    {
    if (x == HUGE_VAL)
    return(1);
    if (x == -HUGE_VAL)
    return(-1);
    return(0);
    }
    #endif /* aixgcc */

    I've attached a little test program, please check it out. It works on my
    Linux AND the AIX box - but Linux does not need it - it has defined
    'isinf'.

    Cheers,
    Gerhard

    PS: I promised to solve that problem, a long, long time ago (shame on me)!

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    - ---1463809280-1334800097-875177029=:1218
    Content-Type: TEXT/PLAIN; charset=US-ASCII; name="testfloat.c"
    Content-Transfer-Encoding: BASE64
    Content-ID: <Pine.LNX.3.95.970925104349.1218B@server>
    Content-Description: testfloat.c

    I2luY2x1ZGUgPG1hdGguaD4NCg0KaW50IGlzaW5mKGRvdWJsZSB4KQ0Kew0K
    CXByaW50ZigiTG9jYWwgZnVuY3Rpb24gaXNpbmYoKSBjYWxsZWQgd2l0aCAl
    Zi4uLiIseCk7DQoJaWYgKHggPT0gSFVHRV9WQUwpDQoJCXJldHVybigxKTsN
    CglpZiAoeCA9PSAtSFVHRV9WQUwpDQoJCXJldHVybigtMSk7DQoJcmV0dXJu
    KDApOw0KfQ0KDQptYWluKCkNCnsNCglpbnQgaW5mOw0KCWRvdWJsZSBwaW5m
    LCBtaW5mLCBydmFsOw0KCXBpbmY9IDEuMC8wLjA7DQoJbWluZj0tMS4wLzAu
    MDsNCglydmFsPSAxLjAvMi4wOw0KDQoJaW5mPWlzaW5mKHBpbmYpOw0KCXN3
    aXRjaCAoaW5mKSB7DQoJCWNhc2UgIDE6IHByaW50ZigiV293IC0gcGluZiBp
    cyBpbmZpbml0ZSAtIHBvc2l0aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0x
    OiBwcmludGYoIldvdyAtIHBpbmYgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFc
    biIpOyBicmVhazsNCgkJZGVmYXVsdDogcHJpbnRmKCJwaW5mIGlzICVmLlxu
    IixwaW5mKTsNCgl9DQoJaW5mPWlzaW5mKG1pbmYpOw0KCXN3aXRjaCAoaW5m
    KSB7DQoJCWNhc2UgIDE6IHByaW50ZigiV293IC0gbWluZiBpcyBpbmZpbml0
    ZSAtIHBvc2l0aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0xOiBwcmludGYo
    IldvdyAtIG1pbmYgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFcbiIpOyBicmVh
    azsNCgkJZGVmYXVsdDogcHJpbnRmKCJtaW5mIGlzICVmLlxuIixtaW5mKTsN
    Cgl9DQoJaW5mPWlzaW5mKHJ2YWwpOw0KCXN3aXRjaCAoaW5mKSB7DQoJCWNh
    c2UgIDE6IHByaW50ZigiV293IC0gcnZhbCBpcyBpbmZpbml0ZSAtIHBvc2l0
    aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0xOiBwcmludGYoIldvdyAtIHJ2
    YWwgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFcbiIpOyBicmVhazsNCgkJZGVm
    YXVsdDogcHJpbnRmKCJydmFsIGlzICVmLlxuIixydmFsKTsNCgl9DQp9DQo=
    - ---1463809280-1334800097-875177029=:1218--

    ------------------------------
  • Bruce Momjian at Sep 25, 1997 at 3:55 pm
    This looks great. Please submit a patch for float.c.

    You can use __GNUC__ as a flag to enable it. See tools/ccsym for a list
    of compiler defines. You can edit the file to use whatever compiler you
    want to test.
    This message is in MIME format. The first part should be readable text,
    while the remaining parts are likely unreadable without MIME-aware tools.
    Send mail to mime@docserver.cac.washington.edu for more info.

    ---1463809280-1334800097-875177029=:1218
    Content-Type: TEXT/PLAIN; charset=US-ASCII
    On Wed, 24 Sep 1997, Bruce Momjian wrote:

    On Wed, 24 Sep 1997, D. Jay Newman wrote:

    Bingo! I just tracked down the gcc float.h, and you are correct.

    I don't know how to deal with this, though.
    ... CUT ...
    I'm thinking of that problem very long - and I only know the (ugly?) way:
    Add the entries:
    -DFP_PLUS_INF=4 -DFP_MINUS_INF=5
    to the compiler flag switches in the aix-gcc template.
    Or, we can add that into float.h if gcc if using aix, and gcc is being
    defined, or you could try:

    #include </usr/include/float.h>

    in the aix-specific code in float.c.

    Does that work? It over-rides the search for includes.
    No very good idea, it produces many double-defined warnings.

    My suggestion (I think, the current implementation does not produce
    correct code with gcc?) is to implement a aixgcc version of isinf - it'll
    be not the only one ;-).

    Example:
    #if defined(aixgcc)
    static int isinf(double x)
    {
    if (x == HUGE_VAL)
    return(1);
    if (x == -HUGE_VAL)
    return(-1);
    return(0);
    }
    #endif /* aixgcc */

    I've attached a little test program, please check it out. It works on my
    Linux AND the AIX box - but Linux does not need it - it has defined
    'isinf'.

    Cheers,
    Gerhard

    PS: I promised to solve that problem, a long, long time ago (shame on me)!

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    ---1463809280-1334800097-875177029=:1218
    Content-Type: TEXT/PLAIN; charset=US-ASCII; name="testfloat.c"
    Content-Transfer-Encoding: BASE64
    Content-ID: <Pine.LNX.3.95.970925104349.1218B@server>
    Content-Description: testfloat.c

    I2luY2x1ZGUgPG1hdGguaD4NCg0KaW50IGlzaW5mKGRvdWJsZSB4KQ0Kew0K
    CXByaW50ZigiTG9jYWwgZnVuY3Rpb24gaXNpbmYoKSBjYWxsZWQgd2l0aCAl
    Zi4uLiIseCk7DQoJaWYgKHggPT0gSFVHRV9WQUwpDQoJCXJldHVybigxKTsN
    CglpZiAoeCA9PSAtSFVHRV9WQUwpDQoJCXJldHVybigtMSk7DQoJcmV0dXJu
    KDApOw0KfQ0KDQptYWluKCkNCnsNCglpbnQgaW5mOw0KCWRvdWJsZSBwaW5m
    LCBtaW5mLCBydmFsOw0KCXBpbmY9IDEuMC8wLjA7DQoJbWluZj0tMS4wLzAu
    MDsNCglydmFsPSAxLjAvMi4wOw0KDQoJaW5mPWlzaW5mKHBpbmYpOw0KCXN3
    aXRjaCAoaW5mKSB7DQoJCWNhc2UgIDE6IHByaW50ZigiV293IC0gcGluZiBp
    cyBpbmZpbml0ZSAtIHBvc2l0aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0x
    OiBwcmludGYoIldvdyAtIHBpbmYgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFc
    biIpOyBicmVhazsNCgkJZGVmYXVsdDogcHJpbnRmKCJwaW5mIGlzICVmLlxu
    IixwaW5mKTsNCgl9DQoJaW5mPWlzaW5mKG1pbmYpOw0KCXN3aXRjaCAoaW5m
    KSB7DQoJCWNhc2UgIDE6IHByaW50ZigiV293IC0gbWluZiBpcyBpbmZpbml0
    ZSAtIHBvc2l0aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0xOiBwcmludGYo
    IldvdyAtIG1pbmYgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFcbiIpOyBicmVh
    azsNCgkJZGVmYXVsdDogcHJpbnRmKCJtaW5mIGlzICVmLlxuIixtaW5mKTsN
    Cgl9DQoJaW5mPWlzaW5mKHJ2YWwpOw0KCXN3aXRjaCAoaW5mKSB7DQoJCWNh
    c2UgIDE6IHByaW50ZigiV293IC0gcnZhbCBpcyBpbmZpbml0ZSAtIHBvc2l0
    aXZlIVxuIik7IGJyZWFrOw0KCQljYXNlIC0xOiBwcmludGYoIldvdyAtIHJ2
    YWwgaXMgaW5maW5pdGUgLSBuZWdhdGl2ZSFcbiIpOyBicmVhazsNCgkJZGVm
    YXVsdDogcHJpbnRmKCJydmFsIGlzICVmLlxuIixydmFsKTsNCgl9DQp9DQo=
    ---1463809280-1334800097-875177029=:1218--

    - --
    Bruce Momjian
    maillist@candle.pha.pa.us

    ------------------------------
  • Gerhard Reithofer at Sep 25, 1997 at 6:12 pm
    This message is in MIME format. The first part should be readable text,
    while the remaining parts are likely unreadable without MIME-aware tools.
    Send mail to mime@docserver.cac.washington.edu for more info.

    - ---1463809280-442363287-875211123=:2495
    Content-Type: TEXT/PLAIN; charset=US-ASCII
    On Thu, 25 Sep 1997, Bruce Momjian wrote:

    This looks great. Please submit a patch for float.c.

    You can use __GNUC__ as a flag to enable it. See tools/ccsym for a list
    of compiler defines. You can edit the file to use whatever compiler you
    want to test.
    ... CUT ...
    Example:
    #if defined(aixgcc)
    static int isinf(double x)
    {
    if (x == HUGE_VAL)
    return(1);
    if (x == -HUGE_VAL)
    return(-1);
    return(0);
    }
    #endif /* aixgcc */
    OK, here it is for V6.1!!!
    It seems to compile...

    I do not use V6.2, but I remembered the "old" problem.
    Please check if it works for 6.2b10!

    My postgres versions are in production and it's very hard to interrupt
    the work for an update (dump, recompile ... test,test,test...).

    I hope it helps even though,
    Gerhard

    +------------------+ +--- gerhardr@tech-edv.co.at ----+
    T.B.Reithofer \ Gerhard | Technical Sofware Developement |
    Staatsbahnstr. 100 \ Reithofer | CAD/CAM/CAE/CAQ/CAP +----------+
    A-2136 Laa/Thaya +----------+ Mechanical CAD +----+
    Tel +43-2522/8726, Fax +43-2522/87268 +---------+
    +---------------------------------------+

    - ---1463809280-442363287-875211123=:2495
    Content-Type: TEXT/PLAIN; charset=US-ASCII; name="float.c.patch"
    Content-Transfer-Encoding: BASE64
    Content-ID: <Pine.LNX.3.95.970925201203.2495C@server>
    Content-Description: float.c.patch

    LS0tIGJhY2tlbmQvdXRpbHMvYWR0L2Zsb2F0LmMub3JpZwlUdWUgSnVsICAx
    IDIwOjM4OjAwIDE5OTcNCisrKyBiYWNrZW5kL3V0aWxzL2FkdC9mbG9hdC5j
    CVRodSBTZXAgMjUgMTk6MTQ6NTAgMTk5Nw0KQEAgLTE0MDEsNiArMTQwMSwx
    OSBAQA0KICNpZm5kZWYgSEFWRV9JU0lORg0KIA0KICNpZiBkZWZpbmVkKGFp
    eCkNCisvKiBUaGUgZ2NjIGRvZXNuJ3Qgc3VwcG9ydCBpc2luZigpICh3aXRo
    b3V0IGxpYmdjYz8pIHNvIHdlIA0KKyAqIGhhdmUgdG8gZG8gaXQgLSBHZXJo
    YXJkIFJlaXRvZmVyIA0KKyAqLw0KKyNpZmRlZiBfX0dOVUNfXw0KK3N0YXRp
    YyBpbnQgaXNpbmYoZG91YmxlIHgpDQorew0KKyAgICBpZiAoeCA9PSBIVUdF
    X1ZBTCkNCisgICAgICAgIHJldHVybigxKTsNCisgICAgaWYgKHggPT0gLUhV
    R0VfVkFMKQ0KKyAgICAgICAgcmV0dXJuKC0xKTsNCisgICAgcmV0dXJuKDAp
    Ow0KK30NCisjZWxzZSAvKiBfX0dOVUNfXyAqLw0KICNpZmRlZiBDTEFTU19D
    T05GTElDVA0KIC8qIHdlIHdhbnQgdGhlIG1hdGggc3ltYm9sICovDQogI3Vu
    ZGVmIGNsYXNzDQpAQCAtMTQxNiw2ICsxNDI5LDcgQEANCiAJcmV0dXJuKC0x
    KTsNCiAgICAgcmV0dXJuKDApOw0KIH0NCisjZW5kaWYgLyogX19HTlVDX18g
    Ki8NCiAjZW5kaWYgLyogYWl4ICovDQogDQogI2lmIGRlZmluZWQodWx0cml4
    NCkNCg==
    - ---1463809280-442363287-875211123=:2495--

    ------------------------------

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
grouppgsql-hackers @
categoriespostgresql
postedSep 23, '97 at 8:44p
activeSep 25, '97 at 6:12p
posts16
users5
websitepostgresql.org...
irc#postgresql

People

Translate

site design / logo © 2021 Grokbase