SEO

The company I work for is in the process of transitioning from SGI Unix
boxes to HP-Ux Itainium machines (HP-UX 11.23i) and I need to port several
GNU software tools to the new Itainium. So far, I have not been able to
successfully build and install any of the GNU software that I have tried.
As a **very** newbie Itainium user, I'm more than a little lost as to how
to proceed. Is it even possible to build GNU software on an Itainium
machine and if so are there any tricks, option overrides, or patches that
are Itainium (or maybe HP-UX) specific ? I've built a rair amount of GNU
software in the past, on various machines, and never had a significant
problem so I suspect there is something specific about setting up the
configure process for the Itainium.



--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

breyfogle <breyfogle@aol.com> wrote:
> The company I work for is in the process of transitioning from SGI
> Unix boxes to HP-Ux Itainium machines (HP-UX 11.23i) and I need to

There is 11iv2, which corresponds to an HP-UX 11.23 kernel, but there
isn't any 11.23i

> port several GNU software tools to the new Itainium. So far, I have
> not been able to successfully build and install any of the GNU
> software that I have tried. As a **very** newbie Itainium user, I'm

Have you installed a compiler? While there is a "bundled" compiler on
every HP-UX system, it is meant just for regenning the kernel and not
for actually compiling applications. To that end, one should install
a "real" compiler - either gcc or preferably (IMO, but that differs
from others) the HP ANSI C/C++ compiler.

> more than a little lost as to how to proceed. Is it even possible
> to build GNU software on an Itainium machine and if so are there any
> tricks, option overrides, or patches that are Itainium (or maybe
> HP-UX) specific ? I've built a rair amount of GNU software in the
> past, on various machines, and never had a significant problem so I
> suspect there is something specific about setting up the configure
> process for the Itainium.

Without knowledge of the specific errors you are encountering, about
the best anyone can suggest presently would be to point you at the
HP-UX porting archives:

http://hpux.connect.org.uk/

and its mirrors. They contain a considerable quantity of freely
available source applications ported to HP-UX. In addition, some of
what you might want may be in the HP Internet Express bundles, which
you should be able to find under:

http://software.hp.com/

rick jones

BTW, I suspect the issues are with getting things setup under HP-UX
rather than the hardware on which it is running.

--
portable adj, code that compiles under more than one compiler
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

On Tue, 30 Jan 2007 11:49:15 -0800, Rick Jones <rick.jones2@hp.com> wrote:

> breyfogle <breyfogle@aol.com> wrote:
>> The company I work for is in the process of transitioning from SGI
>> Unix boxes to HP-Ux Itainium machines (HP-UX 11.23i) and I need to
>
> There is 11iv2, which corresponds to an HP-UX 11.23 kernel, but there
> isn't any 11.23i
>
>> port several GNU software tools to the new Itainium. So far, I have
>> not been able to successfully build and install any of the GNU
>> software that I have tried. As a **very** newbie Itainium user, I'm
>
> Have you installed a compiler? While there is a "bundled" compiler on
> every HP-UX system, it is meant just for regenning the kernel and not
> for actually compiling applications. To that end, one should install
> a "real" compiler - either gcc or preferably (IMO, but that differs
> from others) the HP ANSI C/C++ compiler.
>
>> more than a little lost as to how to proceed. Is it even possible
>> to build GNU software on an Itainium machine and if so are there any
>> tricks, option overrides, or patches that are Itainium (or maybe
>> HP-UX) specific ? I've built a rair amount of GNU software in the
>> past, on various machines, and never had a significant problem so I
>> suspect there is something specific about setting up the configure
>> process for the Itainium.
>
> Without knowledge of the specific errors you are encountering, about
> the best anyone can suggest presently would be to point you at the
> HP-UX porting archives:
>
> http://hpux.connect.org.uk/
>
> and its mirrors. They contain a considerable quantity of freely
> available source applications ported to HP-UX. In addition, some of
> what you might want may be in the HP Internet Express bundles, which
> you should be able to find under:
>
> http://software.hp.com/
>
> rick jones
>
> BTW, I suspect the issues are with getting things setup under HP-UX
> rather than the hardware on which it is running.
>
Thanks for the reply. The new HP has the full install of HP's software
development tools, F90, ANSI C, C++.

I've been to the "hpux.connect.org.uk" site looking for info without
finding much of anything related to Itanium packages. That's actually the
site where I downloaded the source-code packages I'm working with.

As for build errors, It's hard for me right now to know what is important
and what isn't. When I first tried to build GNU a2ps, the configure
script failed because it could not identify a proper host. Setting the
host option to something the script would accept, moved the process a bit
further until a series of compile errors popped up indicating missing
header files (undefined typedefs mainly). Patching those (which should
not have been necessary) got the process to the link step which seemed to
complain about both missing shared libraries and unresolved references.

I'm guessing that the configure script only knows about standard PA-RISC
installations and an Itanium installation is enough different to confuse
the process. I


--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

breyfogle <breyfogle@aol.com> wrote:

[Kudos for not top posting, will say though that trimming previous
text is generally a good thing since almost all news readers these
days will thread posts correctly... ]

> Thanks for the reply. The new HP has the full install of HP's
> software development tools, F90, ANSI C, C++.

Good.

> I've been to the "hpux.connect.org.uk" site looking for info without
> finding much of anything related to Itanium packages. That's
> actually the site where I downloaded the source-code packages I'm
> working with.

These days they build both PA-RISC and Itanium2 versions of what they
work-on/refresh. Taking your a2ps example from below I can see where
you wouldn't think there is much Itanium on the site:

http://hpux.connect.org.uk/hppd/hpux...ipt/a2ps-4.13/

that package on their site goes back to 2000. Looking at some other
packages - say the latest they show on the main page:

http://hpux.connect.org.uk/hppd/hpux...ies/gd-2.0.33/

You will see both PA-RISC and Itanium2.

If there is a package on hpux.connect.org.uk which doesn't have an
Itanium version, I think the best thing to do is to go ahead and
contact the librarian via the Contact Us links scattered hither and
yon:

hpux@connect.org.uk

> As for build errors, It's hard for me right now to know what is
> important and what isn't.

Cut and paste to posts here would be one possibility.

> When I first tried to build GNU a2ps, the configure script failed
> because it could not identify a proper host. Setting the host option
> to something the script would accept, moved the process a bit
> further until a series of compile errors popped up indicating
> missing header files (undefined typedefs mainly). Patching those
> (which should not have been necessary) got the process to the link
> step which seemed to complain about both missing shared libraries
> and unresolved references.

Given the age of connect.org.uk's port of a2ps it might be worthwhile
to find something newer as a starting point. Or asking the
connect.org.uk librarian to. Just guessing, but the missing libraries
were probably prereqs for things that do not ship with HP-UX.

The connect.org.uk folks may not have been as good about listing
build/runtime dependencies on those older packages. On the newer ones
they seem to be pretty good about it, which should make life easier.

> I'm guessing that the configure script only knows about standard
> PA-RISC installations and an Itanium installation is enough
> different to confuse the process.

In the case of a2ps, if there were indeed something looking at the
machine type, that port is old enough it wouldn't have groked ia64
coming-out of uname.

rick jones
--
oxymoron n, Hummer H2 with California Save Our Coasts and Oceans plates
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

On Tue, 30 Jan 2007 15:18:22 -0800, Rick Jones <rick.jones2@hp.com> wrote:

> breyfogle <breyfogle@aol.com> wrote:
>
> [Kudos for not top posting, will say though that trimming previous
> text is generally a good thing since almost all news readers these
> days will thread posts correctly... ]
>
>> Thanks for the reply. The new HP has the full install of HP's
>> software development tools, F90, ANSI C, C++.
>
> Good.
>
>> I've been to the "hpux.connect.org.uk" site looking for info without
>> finding much of anything related to Itanium packages. That's
>> actually the site where I downloaded the source-code packages I'm
>> working with.
>
> These days they build both PA-RISC and Itanium2 versions of what they
> work-on/refresh. Taking your a2ps example from below I can see where
> you wouldn't think there is much Itanium on the site:
>
> http://hpux.connect.org.uk/hppd/hpux...ipt/a2ps-4.13/
>
> that package on their site goes back to 2000. Looking at some other
> packages - say the latest they show on the main page:
>
> http://hpux.connect.org.uk/hppd/hpux...ies/gd-2.0.33/
>
> You will see both PA-RISC and Itanium2.
>
> If there is a package on hpux.connect.org.uk which doesn't have an
> Itanium version, I think the best thing to do is to go ahead and
> contact the librarian via the Contact Us links scattered hither and
> yon:
>
> hpux@connect.org.uk
>
>> As for build errors, It's hard for me right now to know what is
>> important and what isn't.
>
> Cut and paste to posts here would be one possibility.
>
>> When I first tried to build GNU a2ps, the configure script failed
>> because it could not identify a proper host. Setting the host option
>> to something the script would accept, moved the process a bit
>> further until a series of compile errors popped up indicating
>> missing header files (undefined typedefs mainly). Patching those
>> (which should not have been necessary) got the process to the link
>> step which seemed to complain about both missing shared libraries
>> and unresolved references.
>
> Given the age of connect.org.uk's port of a2ps it might be worthwhile
> to find something newer as a starting point. Or asking the
> connect.org.uk librarian to. Just guessing, but the missing libraries
> were probably prereqs for things that do not ship with HP-UX.
>
> The connect.org.uk folks may not have been as good about listing
> build/runtime dependencies on those older packages. On the newer ones
> they seem to be pretty good about it, which should make life easier.
>
>> I'm guessing that the configure script only knows about standard
>> PA-RISC installations and an Itanium installation is enough
>> different to confuse the process.
>
> In the case of a2ps, if there were indeed something looking at the
> machine type, that port is old enough it wouldn't have groked ia64
> coming-out of uname.
>
> rick jones

....Made a bit of progress by building a simplier package (enscript) that
does similar things. I can get enscript build with only the following
warning messages. Something is not right WRT lex (I think). Although the
enscript build completes, when it is run, I can get fatal run-time errors
about these same undefined symbols. Any Ideas ?



cc -g -o states main.o utils.o prims.o process.o gram.o lex.o
.../afmlib/libafm.a ../compat/libcompat.a -ll

ld: (Warning) Unsatisfied symbol "yylenguc" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyinput" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyunput" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyolsp" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyoutput" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yytextuc" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyextra" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyfnd" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyprevious" in file
/usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yylsp" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yylstate" in file /usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yytextarr" in file
/usr/lib/hpux32/libl.so
ld: (Warning) Unsatisfied symbol "yyback" in file /usr/lib/hpux32/libl.so
13 warnings.

--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

breyfogle <breyfogle@aol.com> wrote:

[ trim, please ]

> ...Made a bit of progress by building a simplier package (enscript)
> that does similar things. I can get enscript build with only the
> following warning messages. Something is not right WRT lex (I
> think). Although the enscript build completes, when it is run, I
> can get fatal run-time errors about these same undefined symbols.
> Any Ideas ?

> cc -g -o states main.o utils.o prims.o process.o gram.o lex.o
> ../afmlib/libafm.a ../compat/libcompat.a -ll

> ld: (Warning) Unsatisfied symbol "yylenguc" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyinput" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyunput" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyolsp" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyoutput" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yytextuc" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyextra" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyfnd" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyprevious" in file
> /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yylsp" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yylstate" in file /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yytextarr" in file
> /usr/lib/hpux32/libl.so
> ld: (Warning) Unsatisfied symbol "yyback" in file /usr/lib/hpux32/libl.so
> 13 warnings.


Being woefully ilinfomred about lex (flex) and yacc (bison), I'd be
inclined to think that indeed, something was amis - the build did
indeed use HP's lex and not a flex you installed?

Was there any use of yacc? ISTR that "yy" routines may be yacc
routines - perhaps a -ly is needed as well?

rick jones
--
The computing industry isn't as much a game of "Follow The Leader" as
it is one of "Ring Around the Rosy" or perhaps "Duck Duck Goose."
- Rick Jones
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

On Wed, 31 Jan 2007 10:19:06 -0800, Rick Jones <rick.jones2@hp.com> wrote:

> breyfogle <breyfogle@aol.com> wrote:
>
> [ trim, please ]
>
>> ...Made a bit of progress by building a simplier package (enscript)
>> that does similar things. I can get enscript build with only the
>> following warning messages. Something is not right WRT lex (I
>> think). Although the enscript build completes, when it is run, I
>> can get fatal run-time errors about these same undefined symbols.
>> Any Ideas ?
>
>> cc -g -o states main.o utils.o prims.o process.o gram.o lex.o
>> ../afmlib/libafm.a ../compat/libcompat.a -ll
>
>> ld: (Warning) Unsatisfied symbol "yylenguc" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyinput" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyunput" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyolsp" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyoutput" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yytextuc" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyextra" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyfnd" in file /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyprevious" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yylsp" in file /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yylstate" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yytextarr" in file
>> /usr/lib/hpux32/libl.so
>> ld: (Warning) Unsatisfied symbol "yyback" in file
>> /usr/lib/hpux32/libl.so
>> 13 warnings.
>
>
> Being woefully ilinfomred about lex (flex) and yacc (bison), I'd be
> inclined to think that indeed, something was amis - the build did
> indeed use HP's lex and not a flex you installed?
>
> Was there any use of yacc? ISTR that "yy" routines may be yacc
> routines - perhaps a -ly is needed as well?
>
> rick jones

There is actually quite an extensive e-trail for this same problem dating
back several years that seems to implicate something about HP's ia64 lex
implementation. One thread I found was typical:

http://sources.redhat.com/ml/binutil.../msg00337.html

Unfortunately, I have not found any hint of a solution. Building GNU flex
might work but that also requires GNU Bison and maybe gcc, gmake etc....

--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

breyfogle <breyfogle@aol.com> wrote:
> There is actually quite an extensive e-trail for this same problem
> dating back several years that seems to implicate something about
> HP's ia64 lex implementation. One thread I found was typical:

> http://sources.redhat.com/ml/binutil.../msg00337.html

> Unfortunately, I have not found any hint of a solution.

It sounds like their workaround was listed there:

I worked around the problem by nulling out LEXLIB before
running configure:

export LEXLIB=" "
/.../.../configure ...
make
make install

This causes the configure script to skip the test for LEXLIB
and just use a blank value rather than the normal value of "-ll".

'ar' and the other binutils programs don't actually need "-ll",
because they carefully define their own yywrap().

Did that have any effect in your situation?

> Building GNU flex might work but that also requires GNU Bison and
> maybe gcc, gmake etc....

http://hpux.connect.org.uk/hppd/hpux/Gnu/flex-2.5.4a/

shows an 11.23 Itanium2 version, with _runtime_ dependency on bison:

http://hpux.connect.org.uk/hppd/hpux/Gnu/bison-2.3/

which shows no other run-time dependencies. They both show a
build-time dependency on make:

http://hpux.connect.org.uk/hppd/hpux/Gnu/make-3.81/

all of which seem to be ports from 2006.

rick jones
--
firebug n, the idiot who tosses a lit cigarette out his car window
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

Rick Jones wrote:
> breyfogle <breyfogle@aol.com> wrote:
>> There is actually quite an extensive e-trail for this same problem
>> dating back several years that seems to implicate something about
>> HP's ia64 lex implementation. One thread I found was typical:
>
>> http://sources.redhat.com/ml/binutil.../msg00337.html
>
>> Unfortunately, I have not found any hint of a solution.
>
> It sounds like their workaround was listed there:
>
> I worked around the problem by nulling out LEXLIB before
> running configure:
>
> export LEXLIB=" "
> /.../.../configure ...
> make
> make install
>
> This causes the configure script to skip the test for LEXLIB
> and just use a blank value rather than the normal value of "-ll".
>
> 'ar' and the other binutils programs don't actually need "-ll",
> because they carefully define their own yywrap().
>
> Did that have any effect in your situation?
>
>> Building GNU flex might work but that also requires GNU Bison and
>> maybe gcc, gmake etc....
>
> http://hpux.connect.org.uk/hppd/hpux/Gnu/flex-2.5.4a/
>
> shows an 11.23 Itanium2 version, with _runtime_ dependency on bison:
>
> http://hpux.connect.org.uk/hppd/hpux/Gnu/bison-2.3/
>
> which shows no other run-time dependencies. They both show a
> build-time dependency on make:
>
> http://hpux.connect.org.uk/hppd/hpux/Gnu/make-3.81/
>
> all of which seem to be ports from 2006.
>
> rick jones

Unfortunately, the packages I am trying to build actually use the
missing lex functions. If I link without LEXLIB the build completes but
the application crashes at runtime.

breyfogle <breyfogle@aol.com> wrote:
> Unfortunately, the packages I am trying to build actually use the
> missing lex functions. If I link without LEXLIB the build completes
> but the application crashes at runtime.

Well, I'm tapped-out - it might be good to try email to
hpux-devtools@cxx.cup.hp.com or perhaps cxx-dev@cxx.cup.hp.com - both
lists are maintained via email to majordomo@cxx.cup.hp.com.

rick jones
--
Wisdom Teeth are impacted, people are affected by the effects of events.
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...