Ruby on HP-UX

[Tim Nordloh]

------=_Part_7571_27494188.1142974070555
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

I have read a couple of posts regarding compiling Ruby on HP-UX 11i, and
I'm hoping for a little more insight on the exact requirements of doing tha=
t
type of install. Basically, the posts I have read haven't been
understandable by someone like me. an earlier post referred to "disable
ipv6 and wide-getaddrinfo so it uses Ruby's built-in getaddrinfo()
instead." I have no idea how to do that, I'd love an example command or
something. I am hoping that I can create a step-by-step procedure for
installation on any HP-UX server, for people like me, where the high-level
instructions are not helpful. Without further ado, here is where I've
managed to get so far.

1. Download the 1.8.4 tar.gz file
2. run ./configure
3. run gmake (not make)
4. run gmake all
at this point, I got the following error:
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
ld -b -E -L"../../.." -o ../../../.ext/hppa2.0w-hpux11.11/digest/md5.sl
md5init.o md5ossl.o -lcrypto -ldld -lcrypt -lm -lc
ld: Can't find library: "crypto"
gmake[1]: *** [../../../.ext/hppa2.0w- hpux11.11/digest/md5.sl] Error 1
gmake[1]: Leaving directory `/home/tnordloh/ruby/ruby-1.8.4/ext/digest/md5'
gmake: *** [all] Error 1
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
so, I decided to play with the ext/Setup file mentioned in the README. By
uncommenting 'option nodynamic' I'm able to get a clean gmake, and then
'gmake all', and I have a supposedly working ruby install (at least irb
functions, and a very basic script works). The 'ri' command didn't work
though, so on to that....

so now, I want documentation, which seems to be available by running gmake
install-all
this fails after running for quite some time with the following error:
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
zlib.c:
mcccccccccccccccccc........................................................=
...........................
Generating RI...
/home/tnordloh/ruby/ruby-1.8.4/lib/yaml.rb:87: uninitialized constant
YAML::Syck::Resolver (NameError)
from /home/tnordloh/ruby/ruby-1.8.4/lib/rdoc/ri/ri_descriptions.rb:=
1
from /home/tnordloh/ruby/ruby-1.8.4/lib/rdoc/ri/ri_reader.rb:1
from /home/tnordloh/ruby/ruby-1.8.4/lib/rdoc/generators/ri_generato=
r.rb:46

from /home/tnordloh/ruby/ruby-1.8.4/lib/rdoc/rdoc.rb:250:in
`document'
from ./bin/rdoc:63
gmake: *** [do-install-doc] Error 1

My Ruby-expert friend believes this is because I'm missing some C headers.
Any Ruby install gurus have an idea of what I should try here, or what I'm
missing?

------=_Part_7571_27494188.1142974070555--

 

[James Edward Gray II]

4. run gmake all

Did you mean gmake install-all here?

My Ruby-expert friend

<laughs> I know that refers to me, but boy are you confused.
What I know about the Ruby compile process is limited to: "Works
great on Moc OS X!"

believes this is because I'm missing some C headers.

Actually, I believe Ruby failed to install its C extensions. In a
private message, Tim shared the contents of his extension directory:

# ls /usr/local/lib/ruby/1.8/hppa2.0w-hpux11.11/
bigdecimal.sl defines.h intern.h regex.h util.h
config.h digest missing.h ruby.h version.h
crypto.h digest.sl node.h rubyio.h
curses.sl dln.h rbconfig.rb rubysig.h
dbm.sl env.h re.h st.h

Compare that with my healthy install:

$ ls /usr/local/lib/ruby/1.8/powerpc-darwin8.3.0/
bigdecimal.bundle iconv.bundle rubysig.h
config.h intern.h sdbm.bundle
curses.bundle io socket.bundle
dbm.bundle missing.h st.h
defines.h nkf.bundle stringio.bundle
digest node.h strscan.bundle
digest.bundle openssl.bundle syck.bundle
dl.bundle pty.bundle syslog.bundle
dl.h racc tcltklib.bundle
dlconfig.h rbconfig.rb tkutil.bundle
dln.h re.h util.h
enumerator.bundle readline.bundle version.h
env.h regex.h zlib.bundle
etc.bundle ruby.h
fcntl.bundle rubyio.h

I had him try a few tests. Pure Ruby standard libraries seem to be
just fine, but, as you can see, the C extensions are missing.

James Edward Gray II

 

[Stephen Waits]

but, as you can see, the C extensions are missing.

Yah.. isn't that the rub. In general, the extension build system leaves
a bit to be desired.

Though, if it can find all it needs to build an extension, it should
build it. A few things to try..

* Try editing ext/Setup
* Try adding --with-readline-dir=... --with-openssl-dir=..., to your
/configure

HTH,
Steve

 

[Mauricio Fernandez]

On Mar 21, 2006, at 2:48 PM, Tim Nordloh wrote:
I had him try a few tests. Pure Ruby standard libraries seem to be
just fine, but, as you can see, the C extensions are missing.

That's what he asked for, by disabling shared libs:

He could either uncomment the desired extensions in ext/Setup so they're linked
statically, or skip digest/* (and probably other extensions) to get a clean
build, without disabling dynamic modules altogether.

 

[Tim Nordloh]

------=_Part_357_3026888.1142977869175
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

in what way should I edit /ext/setup? Is there a readme on my options?

Also, I'm assuming the --with-openssl-dir parameter needs to be the
'include' directory? Here's what I tried, and the result.

$ ./configure --with-openssl-dir=3D/usr/local/include/openssl
/configure[88]: conf576.sh: Cannot create the specified file.
/configure[89]: conf576.sh: Cannot create the specified file.
chmod: can't access conf576.sh
/configure[201]: conf576.file: Cannot create the specified file.
/configure[996]: config.log: Cannot create the specified file.

Yah.. isn't that the rub. In general, the extension build system leaves
a bit to be desired.

Though, if it can find all it needs to build an extension, it should
build it. A few things to try..

* Try editing ext/Setup
* Try adding --with-readline-dir=3D... --with-openssl-dir=3D..., to your
./configure

HTH,
Steve

------=_Part_357_3026888.1142977869175--

 

[James Edward Gray II]

That's what he asked for, by disabling shared libs:


He could either uncomment the desired extensions in ext/Setup so
they're linked
statically, or skip digest/* (and probably other extensions) to get
a clean
build, without disabling dynamic modules altogether.

Any idea why digest threw a fit?

James Edward Gray II

 

[Tim Nordloh]

------=_Part_669_23911725.1142978795874
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Looking into it, I might be able to download a newer version, if it's
something I have control over. I'll let you know ASAP

Any idea why digest threw a fit?

James Edward Gray II

------=_Part_669_23911725.1142978795874--

 

[Tim Nordloh]

------=_Part_1245_25143547.1142981272586
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

ok, I uncommented entries in the ext/Setup file one-by-one, and here's what
I ended up with:

root@atcito01 [/home/tnordloh/ruby/ruby-1.8.4]
# cat ext/Setup
#option nodynamic

#Win32API
bigdecimal
curses
dbm
#digest
digest/md5
digest/rmd160
digest/sha1
digest/sha2
#dl
#enumerator
#etc
#fcntl
#gdbm
iconv
#io/wait
#nkf
#pty
openssl
#racc/cparse
#readline
#sdbm
#socket
#stringio
#strscan
#syck
#syslog
#tcltklib
#tk
#win32ole
zlib

I'm back to 'gmake' at this point and it fails at the ruby compile like
so...

making ruby
gmake[1]: Entering directory `/home/tnordloh/ruby/ruby-1.8.4'
gcc -g -O2 -DRUBY_EXPORT -DYYMAXDEPTH=3D300 -I. -I. -oext/extinit.o -c
ext/extinit.c
gcc -g -O2 -DRUBY_EXPORT -DYYMAXDEPTH=3D300 -Wl,-E -L.
-L"/usr/local/include/openssl/lib" -E main.o ext/extinit.o
ext/bigdecimal/bigdecimal.a ext/curses/curses.a ext/dbm/dbm.a
ext/digest/md5/md5.a ext/digest/rmd160/rmd160.a ext/digest/sha1/sha1.a
ext/digest/sha2/sha2.a ext/iconv/iconv.a ext/openssl/openssl.a
ext/zlib/zlib.a -lruby-static -ldld -lcrypt -lm -lcur_colr -ltermcap -lssl
-lcrypto -liconv -lnsl -lz -o ruby
gcc: -E: linker input file unused because linking not done
gcc: main.o: linker input file unused because linking not done
gcc: ext/extinit.o: linker input file unused because linking not done
gcc: ext/bigdecimal/bigdecimal.a: linker input file unused because linking
not done
gcc: ext/curses/curses.a: linker input file unused because linking not done
gcc: ext/dbm/dbm.a: linker input file unused because linking not done
gcc: ext/digest/md5/md5.a: linker input file unused because linking not don=
e
gcc: ext/digest/rmd160/rmd160.a: linker input file unused because linking
not done
gcc: ext/digest/sha1/sha1.a: linker input file unused because linking not
done
gcc: ext/digest/sha2/sha2.a: linker input file unused because linking not
done
gcc: ext/iconv/iconv.a: linker input file unused because linking not done
gcc: ext/openssl/openssl.a: linker input file unused because linking not
done
gcc: ext/zlib/zlib.a: linker input file unused because linking not done
gcc: -lruby-static: linker input file unused because linking not done
gcc: -ldld: linker input file unused because linking not done
gcc: -lcrypt: linker input file unused because linking not done
gcc: -lm: linker input file unused because linking not done
gcc: -lcur_colr: linker input file unused because linking not done
gcc: -ltermcap: linker input file unused because linking not done
gcc: -lssl: linker input file unused because linking not done
gcc: -lcrypto: linker input file unused because linking not done
gcc: -liconv: linker input file unused because linking not done
gcc: -lnsl: linker input file unused because linking not done
gcc: -lz: linker input file unused because linking not done
gmake[1]: Leaving directory `/home/tnordloh/ruby/ruby-1.8.4'

And that really loses me. It looks a lot like the 'gcc' command got
garbled. Any idea what to make of that? I don't see an explicit 'failed'
message, but did it run right? I can't tell.
One thing is for sure, ri still isn't working:

root@atcito01 [/home/tnordloh/ruby/ruby-1.8.4]
# ri Array
/usr/local/lib/ruby/1.8/yaml.rb:87: uninitialized constant
YAML::Syck::Resolver (NameError)
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_descriptions.rb:1
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_reader.rb:1
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_driver.rb:5
from /usr/local/bin/ri:43

Looking into it, I might be able to download a newer version, if it's
something I have control over. I'll let you know ASAP

------=_Part_1245_25143547.1142981272586--

 

[Tim Nordloh]

------=_Part_585_25985389.1143040010478
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

These first couple of lines look like a malformed gcc command to me. Anyon=
e
have an idea of what the line should read?





gcc -g -O2 -DRUBY_EXPORT -DYYMAXDEPTH=3D300 -Wl,-E -L.

-L"/usr/local/include/openssl/lib" -E main.o ext/extinit.o
ext/bigdecimal/bigdecimal.a ext/curses/curses.a ext/dbm/dbm.a
ext/digest/md5/md5.a ext/digest/rmd160/rmd160.a ext/digest/sha1/sha1.a
ext/digest/sha2/sha2.a ext/iconv/iconv.a ext/openssl/openssl.a
ext/zlib/zlib.a -lruby-static -ldld -lcrypt -lm -lcur_colr -ltermcap -lssl
-lcrypto -liconv -lnsl -lz -o ruby

gcc: -E: linker input file unused because linking not done

------=_Part_585_25985389.1143040010478--

 

[Mauricio Fernandez]

These first couple of lines look like a malformed gcc command to me. Anyone
have an idea of what the line should read?

gcc -g -O2 -DRUBY_EXPORT -DYYMAXDEPTH=300 -Wl,-E -L.
-L"/usr/local/include/openssl/lib" -E main.o ext/extinit.o

====
Could you try without this?

ext/bigdecimal/bigdecimal.a ext/curses/curses.a ext/dbm/dbm.a
ext/digest/md5/md5.a ext/digest/rmd160/rmd160.a ext/digest/sha1/sha1.a
ext/digest/sha2/sha2.a ext/iconv/iconv.a ext/openssl/openssl.a
ext/zlib/zlib.a -lruby-static -ldld -lcrypt -lm -lcur_colr -ltermcap -lssl
-lcrypto -liconv -lnsl -lz -o ruby

BTW, you might have to add digest (which digest/md5 and friends depend on)
and syck (in order to get ri to work) to the extension list.

 

[Tim Nordloh]

------=_Part_2827_19261585.1143049403032
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Ok, I hand-modified it and ended up with this....
gcc -g -O2 -DRUBY_EXPORT -DYYMAXDEPTH=3D300 -Wl,-E -L.
main.oext/extinit.o ext/bigdecimal/bigdecimal.a \

ext/curses/curses.a ext/dbm/dbm.a ext/digest/md5/md5.a

ext/digest/rmd160/rmd160.a ext/digest/sha1/sha1.a \

ext/digest/sha2/sha2.a ext/iconv/iconv.a ext/openssl/openssl.a

ext/zlib/zlib.a -lruby-static -ldld -lcrypt \

-lm -lcur_colr -ltermcap -lssl -lcrypto -liconv -lnsl -lz -o ruby

which seemed to run successfully.

So, I went through and removed all references to the
"/usr/local/include/openssl/lib" from all MakeFiles. I still got an error,
so I ended up removing the "EXTLDFLAGS" entry (which contained the extra -E
which was causing the compile failure. I was then able to successfully run
the 'gmake' routine, as well as 'gmake all'.

Thanks, this is the closest I've managed to get, but all that work

"gmake install-all" appeared to successfully run, but I continue to get the
same error when I attempt to run ri:
root@atcito01 [/home/tnordloh/ruby/ruby-1.8.4]
# ri File
/usr/local/lib/ruby/1.8/yaml.rb:87: uninitialized constant
YAML::Syck::Resolver (NameError)
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_descriptions.rb:1
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_reader.rb:1
from /usr/local/lib/ruby/1.8/rdoc/ri/ri_driver.rb:5
from /usr/local/bin/ri:43

So what is the problem?

=3D=3D=3D=3D
Could you try without this?


BTW, you might have to add digest (which digest/md5 and friends depend on= )
and syck (in order to get ri to work) to the extension list.

------=_Part_2827_19261585.1143049403032--

 

[MenTaLguY]

I really think this approach of removing things until it compiles is a mi=
stake; if you did eventually get something that built, you'd just end up =
with a build of Ruby that most likely won't run any useful Ruby scripts.

Your main issue is one of how the libraries Ruby depends on were built or=
installed -- you definitely do want to build them yourself, if you're cu=
rrently trying to use prepackaged versions. Part of the what's going on =
is missing header files, but I have no idea what else is going on. If yo=
u build the libraries yourself with the default options (to the degree po=
ssible), they're a known quantity.

I actually created a /usr/local/ruby where I installed fresh builds of al=
l the required libraries. Note that doing so will required compiling eve=
rything with e.g. -Wl,+b -Wl,/usr/local/ruby/lib though...

Also, see my notes in the earlier post:
http://groups.google.com/group/comp.lang.ruby/browse_thread/thread/fdb97=
e11bcf2be54/d1dc4df9c0a875b8#d1dc4df9c0a875b8

I have to ask, though -- do you really need to do this on _HP-UX_?

For me, it was almost more pain than it was worth -- not because of adjus=
tments required to Ruby itself, but because of quirks of the platform and=
the gymnastics required to build the libraries Ruby depends on in just t=
he right way to make the dynamic linker happy.

-mental

 

[Tim Nordloh]

------=_Part_6261_9099241.1143063959861
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Well, my basic Ruby scripts run ok for my basic needs, and no -- I don't
'need' to run Ruby on HP-UX. I'm the System Admin: I'll just label it as a
'non HP-UX compliant product' and tell any developer that wants to use it t=
o
use Perl for their interpreted language needs

I'll just wait until some bright-eyed kid gets a version on the HP Software
porting archive.

Thanks all, for the attempts....

I really think this approach of removing things until it compiles is a
mistake; if you did eventually get something that built, you'd just end u= p
with a build of Ruby that most likely won't run any useful Ruby scripts.

Your main issue is one of how the libraries Ruby depends on were built or
installed -- you definitely do want to build them yourself, if you're
currently trying to use prepackaged versions. Part of the what's going o= n
is missing header files, but I have no idea what else is going on. If yo= u
build the libraries yourself with the default options (to the degree
possible), they're a known quantity.

I actually created a /usr/local/ruby where I installed fresh builds of al= l
the required libraries. Note that doing so will required compiling
everything with e.g. -Wl,+b -Wl,/usr/local/ruby/lib though...

Also, see my notes in the earlier post:

http://groups.google.com/group/comp.lang.ruby/browse_thread/thread/fdb97e= 11bcf2be54/d1dc4df9c0a875b8#d1dc4df9c0a875b8

I have to ask, though -- do you really need to do this on _HP-UX_?

For me, it was almost more pain than it was worth -- not because of
adjustments required to Ruby itself, but because of quirks of the platfor= m
and the gymnastics required to build the libraries Ruby depends on in jus= t
the right way to make the dynamic linker happy.

-mental

------=_Part_6261_9099241.1143063959861--

 

[MenTaLguY]

I'll just wait until some bright-eyed kid gets a version on the HP
Software porting archive.

I guess that really means getting correctly built versions of all the rel=
event libraries on the porting archive first.

Once that were done, building Ruby proper is trivial aside from the usual=
HP-UX porting tricks like short-circuiting the autoconf test to ignore H=
P-UX's incompletely implemented getaddrinfo().

Does anyone else on the list care enough for me to pursue this?

-mental

 

[Benjohn Barnes]

I guess that really means getting correctly built versions of all
the relevent libraries on the porting archive first.

Once that were done, building Ruby proper is trivial aside from the
usual HP-UX porting tricks like short-circuiting the autoconf test
to ignore HP-UX's incompletely implemented getaddrinfo().

Does anyone else on the list care enough for me to pursue this?

I desperately need a version of Ruby that works on HPUX, and I've
really got very little idea of how to go about getting it to work :
( - I'm from the happy world of opening my iBook, and "oooh, it's pre-
installed!", all this needing to download things, tinker with them,
and build them, is totally alien!

If you can get it to work, and you live in London, I'm happy to
pay in beer

 

[Tim Nordloh]

------=_Part_8483_15826766.1143079471534
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

MenTaLGuY,
I'd love to get Ruby working on HP-UX, but more importantly I would like to
generate a series of steps that can be easily duplicated to install Ruby on
HP-UX anywhere. Of course, this is mostly to benefit me, so it's very easy
to convince me to stop bugging the entire Ruby community about it . As I
mentioned in my original post, the directions out there right now are just
too high-level for me.

I'm willing to help out in whatever way possible in getting a procedure set
up.

If this helps, In the final analysis, I was getting the same failure on all
libraries except one. I couldn't link 'crypto' on most, and 'iconv' on the
odd one. I was able to verify this by purging references to 'crypto' in th=
e
makefiles, at which point everything except iconv successfully linked. Giv=
e
me a hint as to what library I need for this. I actually see crypto.h in
the /usr/local/etc/openssl directory, is there another version I need?

For some reason I still get SYCK errors at this point which, I assume,
means that one of the libraries needs something I messed up by removing
crypto linking and not dynamically linking iconv.

If you have something else for me to try, I'll be glad to.

------=_Part_8483_15826766.1143079471534--

 

[James Edward Gray II]

I'd love to get Ruby working on HP-UX, but more importantly I would
like to
generate a series of steps that can be easily duplicated to install
Ruby on
HP-UX anywhere.

What I believe MenTaLguY was trying to say is that you would need to
sit down and do a bunch of compiling to get it going.

Those C extensions link against other libraries on the system. For
some reason, this isn't working out for your build process. It could
be because you don't have the needed library installed, or just that
your package-manager installs aren't recognized, possibly because of
missing header files.

By pulling those libraries down yourself and building them from
source, you can probably get around at least a large portion of
this. However, that's not likely to boil down to a simple series of
steps.

James Edward Gray II

 

[Tim Nordloh]

------=_Part_9135_10616429.1143083590929
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

At this point, I'm not asking for a lot, just a hint would help, such as,
what library is crypto in (openssl I guess???? I'm attempting to compile
that now, so hopefully my guess is right), and what library is iconv in?
And which one is likely the hangup for YAML+SYCK?

I'm trying my best to not waste anyone's time, and for every response I've
spent at least an hour checking, rechecking deleting and reinstalling from
scratch (honing the mysterious procedure), and I have made several
discoveries on the way. I thank everyone for advice so far and understand
if you all want to give up. After all, I'm not exactly paying anyone for
their time. That's kind of why I keep talking about creating a lower-level
procedure for an HP-UX install, so that the Ruby community will at least be
able to gain some benefit at the end of this thread.

What I believe MenTaLguY was trying to say is that you would need to
sit down and do a bunch of compiling to get it going.

Those C extensions link against other libraries on the system. For
some reason, this isn't working out for your build process. It could
be because you don't have the needed library installed, or just that
your package-manager installs aren't recognized, possibly because of
missing header files.

By pulling those libraries down yourself and building them from
source, you can probably get around at least a large portion of
this. However, that's not likely to boil down to a simple series of
steps.

James Edward Gray II

------=_Part_9135_10616429.1143083590929--

 

[MenTaLguY]

libcrypto is provided by openssl, yes

iconv _should_ be provided by your OS's libc, but as I recall HP-UX's is
present but broken (HP has a distressing habit of implementing things
just enough to pass a naive configure test). You can get a portable
iconv replacement for Ruby to use here:

http://www.gnu.org/software/libiconv/

If I remember correctly, the list of essential libraries is something
like (in no particular order):

- zlib
- libgdbm
- libiconv
- ncurses
- readline
- openssl

I think there's one more I'm forgetting, but I can't find my notes just
now. It'll be pretty obvious if you're still missing one, though.

Make sure that these are all available as either shared libraries, or
(failing that) that they were at least built as position-independent
code. Otherwise the HP-UX dynamic linker will choke at runtime because
it can't relocate the static objects HP's linker blindly imports into
shared libraries.

See my old post that I linked to for more details and other tweaks that
need to be made to specific libraries for HP-UX.

-mental

 

[Tim Nordloh]

------=_Part_9928_979708.1143088563998
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

I'll work on at least openssl, since it is the squeaky wheel at the moment.
I'm having issues with it's makefile at the moment. Incidentally, why are
you refering to libcrypto, when the error is related to crypto? Is crypto
somehow contained in the libcrypto file? I ask because I have a '
libcrypto.sl' file on my system, but no 'crypto.sl', so I sense the
difference may be important to me in some way. I'll paste the error below,
so you don't have to refer back to my original message to the forum...

gmake[1]: Entering directory `/home/tnordloh/ruby/ruby-1.8.4/ext/digest/md5=
'
ld -b -E -L"../../.." -o ../../../.ext/hppa2.0w-hpux11.11/digest/md5.sl
md5init.o md5ossl.o -lcrypto -ldld -lcrypt -lm -lc
*ld: Can't find library: "crypto"*
gmake[1]: *** [../../../.ext/hppa2.0w-hpux11.11/digest/md5.sl] Error 1
gmake[1]: Leaving directory `/home/tnordloh/ruby/ruby-1.8.4/ext/digest/md5'
gmake: *** [all] Error 1


I did read your original message, and I apologize for not understanding it.
It's too high-level for me. For example, when you talk about "disable ipv6
and wide-getaddrinfo so it uses Ruby's built-in getaddrinfo() instead" I as=
k
the question "how?" I don't believe I have IPv6 enabled, so maybe I'm good

libcrypto is provided by openssl, yes

iconv _should_ be provided by your OS's libc, but as I recall HP-UX's is
present but broken (HP has a distressing habit of implementing things
just enough to pass a naive configure test). You can get a portable
iconv replacement for Ruby to use here:

http://www.gnu.org/software/libiconv/

If I remember correctly, the list of essential libraries is something
like (in no particular order):

- zlib
- libgdbm
- libiconv
- ncurses
- readline
- openssl

I think there's one more I'm forgetting, but I can't find my notes just
now. It'll be pretty obvious if you're still missing one, though.

Make sure that these are all available as either shared libraries

how do I make sure of this?
or
(failing that) that they were at least built as position-independent
code. Otherwise the HP-UX dynamic linker will choke at runtime because
it can't relocate the static objects HP's linker blindly imports into
shared libraries.

See my old post that I linked to for more details and other tweaks that
need to be made to specific libraries for HP-UX.

-mental

------=_Part_9928_979708.1143088563998--