[Python-bugs-list] [ python-Bugs-811160 ] HP/UX vs configure

[SourceForge.net]

Bugs item #811160, was opened at 2003-09-23 15:47
Message generated for change (Comment added) made by loewis
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=811160&group_id=5470

Category: Build
Group: None
Status: Open
>Resolution: Accepted
Priority: 5
Submitted By: Anthony Baxter (anthonybaxter)
Assigned to: Nobody/Anonymous (nobody)
Summary: HP/UX vs configure

Initial Comment:
configure is busted when building with HP's compiler.

"cc -Ac" is ancient and broken, "cc -Aa" is too anal to

compile it's own system headers, and "cc -Ae" breaks

autoconf.

The problem is the same one as outlined here:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12100

but in our case, it hits a whole lot of functions.



(basically: AC_CHECK_FUNCS expands to code that ends up

#including limits.h, which ends up including

sys/time.h, which breaks all checks for the various

time related bits.)



The GCC people worked around it with their own version

of the appropriate magic.





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

>Comment By: Martin v. Löwis (loewis)
Date: 2003-09-27 09:40

Message:
Logged In: YES 
user_id=21627

The patch is fine; please apply.

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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-26 20:47

Message:
Logged In: YES 
user_id=29957

I'll also attach a patch for the (generated) configure

script, for those folks without autoconf on their systems.



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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-26 19:14

Message:
Logged In: YES 
user_id=29957

argh. sf tracker strikes again. this time for sure.



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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-26 19:13

Message:
Logged In: YES 
user_id=29957

There's three ways we can fix this:

1. Hand-edit configure after running autoconf (yuk)

2. Run a CVS version of autoconf (yuk)

3. Supply an aclocal.m4 in the same way that the gcc guys

used to fix the problem (a bit yuk).



The aclocal fix would only be needed until the next release

of autoconf is out. I've attached an aclocal.m4 that

includes the GCC fix. Trying it here shows that it generates

a "fixed" configure scipt - I've not yet logged onto the HP

testdrive boxes to confirm it absolutely works, but it looks

like the fixes I did by hand in order to get it to work.



MvL, your thoughts? You seem to be the local autoconf guru... 

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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-23 16:26

Message:
Logged In: YES 
user_id=29957

mwh points out that the snakefarm is having the same problem:

http://www.lysator.liu.se/xenofarm/python/files/792_10/python-testlog.txt

(we discovered and debugged this on one of the

DEC^WDigital^WCompaq^WHP "testdrive" boxes)



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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-23 15:58

Message:
Logged In: YES 
user_id=29957

Oh, and the problem also occurs if you use gcc instead of cc -Ae





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

Comment By: Anthony Baxter (anthonybaxter)
Date: 2003-09-23 15:56

Message:
Logged In: YES 
user_id=29957

the functions that fail (erroneously) are

HAVE_CLOCK HAVE_GETTIMEOFDAY HAVE_MKTIME HAVE_SELECT

HAVE_STRFTIME HAVE_STRPTIME HAVE_UTIMES





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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=811160&group_id=5470