bug-gnu-utils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: building a dynamically linked Perl 5.6.1 on HPUX 11.0


From: Lincoln A. Baxter
Subject: Re: building a dynamically linked Perl 5.6.1 on HPUX 11.0
Date: Thu, 18 Jul 2002 20:17:27 -0400
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.9) Gecko/20020408

Thank you for the suggestion. I will be working on a new version of the README this weekend (Tim Bunce will be releasing a new DBD-Oracle soon), so your message is auspicious. I will add this.
But let me ask you:

You followed the README, but used gcc instead of /opt/softbench/bin/cc and it worked? Can you repeat it? If so can you provide a script or make file that does it? If you can, "you will be helping to save the world".

At the very least, can you send me your output of perl -V?

Thanks, oh.. and I'm glad you found it helpful.... so your're welcome.

With regards,

Lincoln

Michael Schuh wrote:

Greetings all,

I recently (like, 40 minutes ago) installed Perl 5.6.1 on an HPUX 11.0 system.  As my 
goal is to interact with an Oracle database, I followed Lincoln Baxter's README.hpux 
(THANK YOU!) and tried to build a dynamically linked version of Perl.  However, in 
earlier attempts (stretching back over the last week or so) it died in the linking stage, 
admonishing me that a file was created without position independent code and advising me 
that I should recompile using the "+z" flag.  As I was using gcc (which has no 
+z flag) this was very confusing until I noticed that the file to which the message was 
referring was libgdbm, which I had built just last week, and realized that I was using 
HPUX's ld ('cuz the GNU binutils ld won't work on HPUX) and this ld doesn't know about 
gcc.

Well, it was simple enough to fix: go back to the libgdm build and add -fPIC to 
the complier options.  After that, Perl built and loaded just like the 
documentation says it should.

My "bug"?  It would have helped me, I think, if the three sets of documentation were more 
blatant about building everything PIC.  The configure/build process for libgdbm does not have an 
obvious method to specify PIC, and neither the Perl documentation nor Lincoln Baxter's README.hpux 
remind the user that all files in a dynamic build need to be PIC.  Yes, it's "well known" 
(and, if I had been asked, would have said as much...), but a mild reminder might have helped me.

Also, in the Makefile for libgdbm:

30c30
< CFLAGS = -O -fPIC
---

CFLAGS = -O

although this is specific to gcc.

As is often the case in the software biz, just a few characters can make all 
the difference.

Thank you, and thank you very much for various contributions.

--
Mike Schuh









reply via email to

[Prev in Thread] Current Thread [Next in Thread]