Re: 4.2.0-pre-1

Re: 4.2.0-pre-1

[Peter Stephenson]

Ibraheem Umaru-Mohammed wrote:
> I have been unable to build zsh-4.2.0-pre-1, (or zsh-4.1.1) on AIX 5.2. Any
> help will be greatly appreciated. I have included the config.log, and the
> last few lines before failure - should you require any more details let me
> know.
>...
>         /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o init.o ini
> t.c
> "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the en
> d of line.
> "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the
>  end of line.
> "init.c", line 493.5: 1506-275 (S) Unexpected text 'int' encountered.
> make: The error code from the last command is 1.

Thanks for reporting this, although please send messages to
zsh-workers@sunsite.dk, not to me personally.

The code you must be executing looks uncompilable in standard C... not
surprising it didn't work.  Please do let us know about any more
failures, we only have a limited user base for AIX.

pws

Index: Src/init.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/init.c,v
retrieving revision 1.39
diff -u -r1.39 init.c
--- Src/init.c	15 Dec 2003 22:45:29 -0000	1.39
+++ Src/init.c	1 Mar 2004 10:37:40 -0000
@@ -481,6 +481,9 @@
 init_shout(void)
 {
     static char shoutbuf[BUFSIZ];
+#if defined(JOB_CONTROL) && defined(TIOCSETD) && defined(NTTYDISC)
+    int ldisc;
+#endif
 
     if (SHTTY == -1)
     {
@@ -490,8 +493,7 @@
     }
 
 #if defined(JOB_CONTROL) && defined(TIOCSETD) && defined(NTTYDISC)
-    int ldisc = NTTYDISC;
-
+    ldisc = NTTYDISC;
     ioctl(SHTTY, TIOCSETD, (char *)&ldisc);
 #endif
 
-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
The information transmitted is intended only for the person or
entity to which it is addressed and may contain confidential 
and/or privileged material. 
Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by 
persons or entities other than the intended recipient is 
prohibited. 
If you received this in error, please contact the sender and 
delete the material from any computer.
**********************************************************************

Re: 4.2.0-pre-1

[Ibraheem Umaru-Mohammed]

On [01/03/04 10:52 +0000], Peter Stephenson wrote:
> Ibraheem Umaru-Mohammed wrote:
> > I have been unable to build zsh-4.2.0-pre-1, (or zsh-4.1.1) on AIX 5.2. Any
> > help will be greatly appreciated. I have included the config.log, and the
> > last few lines before failure - should you require any more details let me
> > know.
> >...
> >         /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o init.o ini
> > t.c
> > "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the en
> > d of line.
> > "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the
> >  end of line.
> > "init.c", line 493.5: 1506-275 (S) Unexpected text 'int' encountered.
> > make: The error code from the last command is 1.
> 
> Thanks for reporting this, although please send messages to
> zsh-workers@sunsite.dk, not to me personally.
[...]

My apologies - I guess one can still get shy behind the mask of an email.

> The code you must be executing looks uncompilable in standard C... not
> surprising it didn't work.  Please do let us know about any more
> failures, we only have a limited user base for AIX.
> 

This makes sense, because of all the platforms I have access to at work
(Linux, Solaris, HPUX, and AIX), AIX 5.x is the only platform that zsh has 
failed to successfully compile under.

> pws
> 
> Index: Src/init.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/init.c,v
> retrieving revision 1.39
> diff -u -r1.39 init.c
> --- Src/init.c	15 Dec 2003 22:45:29 -0000	1.39
> +++ Src/init.c	1 Mar 2004 10:37:40 -0000
> @@ -481,6 +481,9 @@
>  init_shout(void)
>  {
>      static char shoutbuf[BUFSIZ];
> +#if defined(JOB_CONTROL) && defined(TIOCSETD) && defined(NTTYDISC)
> +    int ldisc;
> +#endif
>  
>      if (SHTTY == -1)
>      {
> @@ -490,8 +493,7 @@
>      }
>  
>  #if defined(JOB_CONTROL) && defined(TIOCSETD) && defined(NTTYDISC)
> -    int ldisc = NTTYDISC;
> -
> +    ldisc = NTTYDISC;
>      ioctl(SHTTY, TIOCSETD, (char *)&ldisc);
>  #endif
>  
[...]

Seem to have gotten further, but still falls over:

[--- build zsh on AIX 5.2 ---]
** NONE **:44% make
        cd . && /bin/sh ./config.status config.h
config.status: creating config.h
config.status: config.h is unchanged
Updated `zsh.mdh'.
Updated `termcap.mdh'.
Updated `terminfo.mdh'.
Updated `cap.mdh'.
Updated `clone.mdh'.
Updated `datetime.mdh'.
Updated `example.mdh'.
Updated `files.mdh'.
Updated `langinfo.mdh'.
Updated `mapfile.mdh'.
Updated `mathfunc.mdh'.
Updated `parameter.mdh'.
Updated `pcre.mdh'.
Updated `socket.mdh'.
Updated `stat.mdh'.
Updated `system.mdh'.
Updated `tcp.mdh'.
Updated `zftp.mdh'.
Updated `zprof.mdh'.
Updated `zpty.mdh'.
Updated `zselect.mdh'.
Updated `zle.mdh'.
Updated `complete.mdh'.
Updated `zutil.mdh'.
Updated `rlimits.mdh'.
Updated `sched.mdh'.
Updated `compctl.mdh'.
Updated `complist.mdh'.
Updated `computil.mdh'.
Updated `deltochar.mdh'.
Updated `zleparameter.mdh'.
        rm -f stamp-modobjs.tmp
Updated `zsh.mdh'.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o builtin.o builtin.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o compat.o compat.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o cond.o cond.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o exec.o exec.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o glob.o glob.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o hashtable.o hashtable.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
"/usr/include/rpcsvc/ypclnt.h", line 99.31: 1506-310 (I) The type "struct dom_binding" was introduced in a parameter list, and will go out of scope at the end of the function declaration or definition.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o hist.o hist.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
`zshpaths.h' is up to date.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o init.o init.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
    1265  1500-010: (W) WARNING in zsh_main: Infinite loop.  Program may not stop.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o input.o input.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
        /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o jobs.o jobs.c
"sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the end of line.
"signames.epro", line 5.74: 1506-186 (W) String literal must be ended before the end of line.
"jobs.c", line 686.43: 1506-276 (S) Syntax error: possible missing ')'?
"jobs.c", line 697.46: 1506-068 (S) Operation between types "char*" and "int" is not allowed.
"jobs.c", line 698.21: 1506-277 (S) Syntax error: possible missing ')' or ','?
"jobs.c", line 698.27: 1506-275 (S) Unexpected text ')' encountered.
"jobs.c", line 769.77: 1506-276 (S) Syntax error: possible missing ')'?
"jobs.c", line 773.88: 1506-277 (S) Syntax error: possible missing ')' or ','?
"jobs.c", line 775.77: 1506-277 (S) Syntax error: possible missing ')' or ','?
make: The error code from the last command is 1.


Stop.
make: The error code from the last command is 2.


Stop.
make: The error code from the last command is 1.


Stop.

[--- build zsh on AIX 5.2 ---]

Cheers,

			--ibraheem

Re: 4.2.0-pre-1

[Peter Stephenson]

Ibraheem Umaru-Mohammed wrote:
> "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the
>  end of line.
> "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before 
> the end of line.
>         /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o jobs.o 
> jobs.c
> "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the
>  end of line.
> "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before 
> the end of line.
> "jobs.c", line 686.43: 1506-276 (S) Syntax error: possible missing ')'?
> "jobs.c", line 697.46: 1506-068 (S) Operation between types "char*" and "int"
>  is not allowed.
> "jobs.c", line 698.21: 1506-277 (S) Syntax error: possible missing ')' or ','
> ?
> "jobs.c", line 698.27: 1506-275 (S) Unexpected text ')' encountered.
> "jobs.c", line 769.77: 1506-276 (S) Syntax error: possible missing ')'?
> "jobs.c", line 773.88: 1506-277 (S) Syntax error: possible missing ')' or ','
> ?
> "jobs.c", line 775.77: 1506-277 (S) Syntax error: possible missing ')' or ','
> ?

It seems to be falling over on the sigmsg(...) macros.

You should have a file in the Src directory where you are building
called sigcount.h containing something like

#define SIGCOUNT	45
#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal")

Is your version similar?  If so, is there something weird in the file
signames.c in the same directory (feel free to send to the file)?

Thanks for the info.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-1

[Ibraheem Umaru-Mohammed]

On [02/03/04 11:15 +0000], Peter Stephenson wrote:
> Ibraheem Umaru-Mohammed wrote:
> > "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the
> >  end of line.
> > "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before 
> > the end of line.
> >         /usr/vac/bin/cc -qlanglvl=ansi -c -I.  -DHAVE_CONFIG_H -O  -o jobs.o 
> > jobs.c
> > "sigcount.h", line 2.74: 1506-186 (W) String literal must be ended before the
> >  end of line.
> > "signames.epro", line 5.74: 1506-186 (W) String literal must be ended before 
> > the end of line.
> > "jobs.c", line 686.43: 1506-276 (S) Syntax error: possible missing ')'?
> > "jobs.c", line 697.46: 1506-068 (S) Operation between types "char*" and "int"
> >  is not allowed.
> > "jobs.c", line 698.21: 1506-277 (S) Syntax error: possible missing ')' or ','
> > ?
> > "jobs.c", line 698.27: 1506-275 (S) Unexpected text ')' encountered.
> > "jobs.c", line 769.77: 1506-276 (S) Syntax error: possible missing ')'?
> > "jobs.c", line 773.88: 1506-277 (S) Syntax error: possible missing ')' or ','
> > ?
> > "jobs.c", line 775.77: 1506-277 (S) Syntax error: possible missing ')' or ','
> > ?
> 
> It seems to be falling over on the sigmsg(...) macros.
> 
> You should have a file in the Src directory where you are building
> called sigcount.h containing something like
> 
> #define SIGCOUNT	45
> #define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal")
> 
> Is your version similar?  If so, is there something weird in the file
> signames.c in the same directory (feel free to send to the file)?
> 
> Thanks for the info.
> 

Hmmmn, sigcount.h seems to have unclosed quotes:

,---- [ sigcount.h ]
| #define SIGCOUNT        59
| #define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal3)
`----

Any utility version you want me to test for? awk? sed? A regular expression
issue?

Cheers,

			--ibraheem

-- 

Re: 4.2.0-pre-1

[Peter Stephenson]

Ibraheem Umaru-Mohammed wrote:
> Hmmmn, sigcount.h seems to have unclosed quotes:
> 
> ,---- [ sigcount.h ]
> | #define SIGCOUNT        59
> | #define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal3)
> `----

Yes, I can believe that's wrong...  Presumably it's also wrong in
signames.c.

The problem must be in signames2.awk which generates this:

    printf "#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)", 34, "unknown signal", 34

My guess is your awk is truncating the input line in some weird way.
(It can't be 80 characters, however, since you wouldn't get the `unknown
signal' at all.)  Can you try altering that line as per the following
patch and see if that helps?

If your awk is playing up, you might find other weirdness, in particular
the `# ifdef USE_SUSPENDED' bit in signames.c.  So even if it looks like
it works it would be worth checking (or posting) the resulting signames.c.
Any fixes ought to be similarly trivial, but I wouldn't like to second
guess what awk is actually getting up to.

Index: Src/signames2.awk
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/signames2.awk,v
retrieving revision 1.1.1.4
diff -u -r1.1.1.4 signames2.awk
--- Src/signames2.awk	15 Feb 2000 16:44:54 -0000	1.1.1.4
+++ Src/signames2.awk	2 Mar 2004 12:05:38 -0000
@@ -63,7 +63,8 @@
     printf "#include %czsh.mdh%c\n", 34, 34
     printf "\n"
     printf "/**/\n"
-    printf "#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)", 34, "unknown signal", 34
+    printf "#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig]"
+    printf " : %c%s%c)", 34, "unknown signal", 34
     printf "\n"
     printf "/**/\n"
     printf "mod_export char *sig_msg[SIGCOUNT+2] = {\n"

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-1

[Ibraheem Umaru-Mohammed]

On [02/03/04 12:14 +0000], Peter Stephenson wrote:
> Ibraheem Umaru-Mohammed wrote:
> > Hmmmn, sigcount.h seems to have unclosed quotes:
> > 
> > ,---- [ sigcount.h ]
> > | #define SIGCOUNT        59
> > | #define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal3)
> > `----
> 
> Yes, I can believe that's wrong...  Presumably it's also wrong in
> signames.c.
> 
> The problem must be in signames2.awk which generates this:
> 
>     printf "#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)", 34, "unknown signal", 34
> 
> My guess is your awk is truncating the input line in some weird way.
> (It can't be 80 characters, however, since you wouldn't get the `unknown
> signal' at all.)  Can you try altering that line as per the following
> patch and see if that helps?
> 
> If your awk is playing up, you might find other weirdness, in particular
> the `# ifdef USE_SUSPENDED' bit in signames.c.  So even if it looks like
> it works it would be worth checking (or posting) the resulting signames.c.
> Any fixes ought to be similarly trivial, but I wouldn't like to second
> guess what awk is actually getting up to.
> 
[...]

I think I've found what the problem is, but have no idea why it is happening.
Take a look at the following:

Linux:

	[ibraheem@sonic:~ ] 42 % uname -a
	Linux sonic 2.6.3-rc2-gentoo #2 SMP Wed Feb 18 09:12:33 GMT 2004 i686
	Pentium III (Coppermine) GenuineIntel GNU/Linux
	[ibraheem@sonic:~ ] 43 %  nawk 'BEGIN {printf "\n\n#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)\n", 34, "unknown signal", 34}'


	#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown
	signal")
	[ibraheem@sonic:~ ] 44 %

AIX 4.3:
	
	[ibraheem@maserati:~ ] 45 % uname -a
	AIX maserati 3 4 0043CDDA4C00
	[ibraheem@maserati:~ ] 46 % nawk 'BEGIN {printf "\n\n#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)\n", 34, "unknown signal", 34}'


	#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal")
	[ibraheem@maserati:~ ] 47 %

AIX 5.2

	ibraheem@md000002s:~ ] 48 % uname -a
	AIX md000002s 2 5 00541F5A4C00
	ibraheem@md000002s:~ ] 49 % nawk 'BEGIN {printf "\n\n#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : %c%s%c)\n", 34, "unknown signal", 34}'


	#define sigmsg(sig) ((sig) <= SIGCOUNT ? sig_msg[sig] : "unknown signal3)
	ibraheem@md000002s: ~] 50 % 

However, it appears that /bin/nawk is the same as /bin/awk, and not a symlink as it should be.
I will attempt to compile up nawk, and use that. 

Cheers,

			--ibraheem.

-- 

4.2.0-pre-1 on HP-UX 11.11

[Paul Ackersviller]

On HP-UX 11.11 I'm getting a couple of configuration problems, it
doesn't sound like these would've gone away on 4.2.0-pre-2.

First one:
cc -c -I.  -DHAVE_CONFIG_H -O  -o builtin.o builtin.c
cc: "prototypes.h", line 106: error 1711: Inconsistent parameter list declaration for "gettimeofday".

I noticed zsh-4.0.8 has
config.h:473:/* Define if you have the gettimeofday function.  */
config.h:474:#define HAVE_GETTIMEOFDAY 1

whereas zsh-4.2.0-pre-1 shows intead
config.h:229:/* Define to 1 if you have the `gettimeofday' function. */
config.h:230:/* #undef HAVE_GETTIMEOFDAY */

and got rid of the problem by defining HAVE_GETTIMEOFDAY.


Second one:
cc: "zselect.c", line 267: error 1634: Missing arguments only allowed on intrinsic calls.

This line is a zerrnam() call, but given that the OS has a select
system call it looks like configure should set me up to compile the
code inside HAVE_SELECT instead.

I haven't been running this version much yet, but I'll let you know
how it goes.

Re: 4.2.0-pre-1 on HP-UX 11.11

[Peter Stephenson]

Paul Ackersviller wrote:
> On HP-UX 11.11 I'm getting a couple of configuration problems, it
> doesn't sound like these would've gone away on 4.2.0-pre-2.

Does config.log explain why gettimeofday() and select() weren't found?
(The manual pages might give some useful info, too.)  Autoconf
2.5x may be stricter.  This really needs fixing.

> Second one:
> cc: "zselect.c", line 267: error 1634: Missing arguments only allowed on intr
> insic calls.

I can at least fix that, but as you say it's just a symptom.

Index: Src/Modules/zselect.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Modules/zselect.c,v
retrieving revision 1.4
diff -u -r1.4 zselect.c
--- Src/Modules/zselect.c	25 Apr 2003 11:19:10 -0000	1.4
+++ Src/Modules/zselect.c	3 Mar 2004 15:16:43 -0000
@@ -265,7 +265,7 @@
 #else
     /* TODO: use poll */
     zerrnam(nam, "your system does not implement the select system call.",
-	    NULL, );
+	    NULL, 0);
     return 2;
 #endif
 }


-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-1 on HP-UX 11.11

[Paul Ackersviller]

On Wed, Mar 03, 2004 at 03:31:48PM +0000, Peter Stephenson wrote:
> Does config.log explain why gettimeofday() and select() weren't found?

It does say the following.  Are there any flags that would get
configure to tell me more?

cc: "configure", line 9234: error 1584: Inconsistent type declaration: "gettimeofday".
cc: "configure", line 9234: error 1584: Inconsistent type declaration: "select".

> (The manual pages might give some useful info, too.)  Autoconf
> 2.5x may be stricter.  This really needs fixing.

I didn't see anything enlightening regarding gettimeofday, but there
is some weirdness with select where a subtly different prototype is
used in HP-UX extended via <time.h> mode in place of <sys/time.h>.

> > Second one:
> > cc: "zselect.c", line 267: error 1634: Missing arguments only allowed on intr
> > insic calls.
> 
> I can at least fix that, but as you say it's just a symptom.

Yes, this compiles and also links with the system curses.  However,
I notice linking against ncurses still fails without the HAVE_SELECT.
Let me know if you'd like any more details on that.

Re: 4.2.0-pre-1 on HP-UX 11.11

[Peter Stephenson]

Paul Ackersviller wrote:
> On Wed, Mar 03, 2004 at 03:31:48PM +0000, Peter Stephenson wrote:
> > Does config.log explain why gettimeofday() and select() weren't found?
> 
> It does say the following.  Are there any flags that would get
> configure to tell me more?
> 
> cc: "configure", line 9234: error 1584: Inconsistent type declaration: "getti
> meofday".
> cc: "configure", line 9234: error 1584: Inconsistent type declaration: "selec
> t".

It should show the failed program, too --- you should have a chunk
beginning:

configure: failed program was:

Without this we don't have enough to go on.

I notice my config.log produces a cryptic message about problems with
gettimeofday on HPUX, but I have no idea what to do about it --- it
gives the impression that there is an adequate workaround.

I could propose alternative test programs for select and gettimeofday,
but without knowing what the problem is it's pointless --- this is
exactly what configure is supposed to be doing.

We desperately need developers (or at least configure experts) with
access to HPUX and AIX.  Otherwise we're stuck with tricks like:

#ifdef __hpux
#undef HAVE_SELECT
#undef HAVE_GETTIMEOFDAY
#define HAVE_SELECT 1
#define HAVE_GETTIMEOFDAY 1
#endif

which is horrible but might at least work.  (If you want to try it,
stick it at the end of Src/system.h and see if it now configures and
builds from scratch.)

By the way, it's possible the same problem exists with HAVE_POLL: it
should use poll() as an alternative to select().

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-1 on HP-UX 11.11

[Paul Ackersviller]

[-- Attachment #1: Type: text/plain, Size: 1682 bytes --]

On Thu, Mar 04, 2004 at 12:21:14PM +0000, Peter Stephenson wrote:
> Paul Ackersviller wrote:
> > cc: "configure", line 9234: error 1584: Inconsistent type declaration: "gettimeofday".
> > cc: "configure", line 9234: error 1584: Inconsistent type declaration: "select".
> 
> It should show the failed program, too --- you should have a chunk
> beginning:
> 
> configure: failed program was:
> 
> Without this we don't have enough to go on.

I've attached the portion in question of my config.log -- I don't
understand why it appears to be trying signatures of char (*)().

> We desperately need developers (or at least configure experts) with
> access to HPUX and AIX.  Otherwise we're stuck with tricks like:

I'm far from a configure expert, but I'd be willing to do what I can.
However I have no AIX access, only HP-UX 11.11 and 11.00 (although the
latter box may not stay around long).

BTW, I tried dynamic-loading some pieces of zsh, which is by default
all configured off on HP-UX.  It looks to work on 11.11.

> #ifdef __hpux
> #undef HAVE_SELECT
> #undef HAVE_GETTIMEOFDAY
> #define HAVE_SELECT 1
> #define HAVE_GETTIMEOFDAY 1
> #endif
> 
> which is horrible but might at least work.  (If you want to try it,
> stick it at the end of Src/system.h and see if it now configures and
> builds from scratch.)

Yes, this does, but when resorting to that wouldn't it be best to make
it version-specific?  Unfortunately I don't see any suitable macros.

> By the way, it's possible the same problem exists with HAVE_POLL: it
> should use poll() as an alternative to select().

No, it seems to be finding poll() alright, although I editted that
portion out of the attached log file.

[-- Attachment #2: config.log --]
[-- Type: text/plain, Size: 7546 bytes --]

configure:9134: checking for gettimeofday
configure:9184: cc -o conftest  -O   conftest.c -lnsl -lcurses -lHcurses -lm  -lc >&5
cc: "configure", line 9234: error 1584: Inconsistent type declaration: "gettimeofday".
cc: "configure", line 9241: warning 604: Pointers are not assignment-compatible.
cc: "configure", line 9250: warning 605: Illegal pointer combination for !=.
configure:9187: $? = 1
configure: failed program was:
| #line 9139 "configure"
| /* confdefs.h.  */
| 
| #define PACKAGE_NAME ""
| #define PACKAGE_TARNAME ""
| #define PACKAGE_VERSION ""
| #define PACKAGE_STRING ""
| #define PACKAGE_BUGREPORT ""
| #define MACHTYPE "hppa2.0w"
| #define VENDOR "hp"
| #define OSTYPE "hpux11.11"
| #define GLOBAL_ZSHENV "/etc/zshenv"
| #define GLOBAL_ZSHRC "/etc/zshrc"
| #define GLOBAL_ZPROFILE "/etc/zprofile"
| #define GLOBAL_ZLOGIN "/etc/zlogin"
| #define GLOBAL_ZLOGOUT "/etc/zlogout"
| #define RESTRICTED_R 1
| #define CONFIG_LOCALE 1
| #define PROTOTYPES 1
| #define HAVE_ALLOCA_H 1
| #define HAVE_ALLOCA 1
| #define HAVE_UNION_INIT 1
| #define HAVE_VARIABLE_LENGTH_ARRAYS 1
| #define HAVE_DIRENT_H 1
| #define STDC_HEADERS 1
| #define TIME_WITH_SYS_TIME 1
| #define HAVE_SYS_WAIT_H 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_SYS_STAT_H 1
| #define HAVE_STDLIB_H 1
| #define HAVE_STRING_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_STRINGS_H 1
| #define HAVE_INTTYPES_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_SYS_TIME_H 1
| #define HAVE_SYS_TIMES_H 1
| #define HAVE_TERMIO_H 1
| #define HAVE_TERMIOS_H 1
| #define HAVE_SYS_PARAM_H 1
| #define HAVE_STRING_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_LIMITS_H 1
| #define HAVE_FCNTL_H 1
| #define HAVE_SYS_UTSNAME_H 1
| #define HAVE_SYS_RESOURCE_H 1
| #define HAVE_LOCALE_H 1
| #define HAVE_ERRNO_H 1
| #define HAVE_STDIO_H 1
| #define HAVE_STDLIB_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_UTMP_H 1
| #define HAVE_UTMPX_H 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_PWD_H 1
| #define HAVE_GRP_H 1
| #define HAVE_POLL_H 1
| #define HAVE_SYS_MMAN_H 1
| #define HAVE_NETINET_IN_SYSTM_H 1
| #define HAVE_LANGINFO_H 1
| #define HAVE_WCHAR_H 1
| #define HAVE_STDDEF_H 1
| #define HAVE_SYS_STROPTS_H 1
| #define HAVE_DLFCN_H 1
| #define HAVE_DL_H 1
| #define HAVE_LIBM 1
| #define HAVE_CURSES_H 1
| #define HAVE_TERM_H 1
| #define RETSIGTYPE void
| #define ZSH_64_BIT_TYPE long long
| #define ZSH_64_BIT_UTYPE unsigned long long
| #define HAVE_STRUCT_TIMEZONE 1
| #define HAVE_STRUCT_UTMP 1
| #define HAVE_STRUCT_UTMPX 1
| #define HAVE_STRUCT_UTMP_UT_HOST 1
| #define HAVE_STRUCT_UTMPX_UT_HOST 1
| #define HAVE_STRUCT_UTMPX_UT_TV 1
| #define HAVE_STRUCT_DIRENT_D_INO 1
| #define HAVE_STRUCT_SOCKADDR_IN6_SIN6_SCOPE_ID 1
| /* end confdefs.h.  */
| /* System header to define __stub macros and hopefully few prototypes,
|     which can conflict with char gettimeofday (); below.
|     Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
|     <limits.h> exists even on freestanding compilers.  */
| #ifdef __STDC__
| # include <limits.h>
| #else
| # include <assert.h>
| #endif
| /* Override any gcc2 internal prototype to avoid an error.  */
| #ifdef __cplusplus
| extern "C"
| {
| #endif
| /* We use char because int might match the return type of a gcc2
|    builtin and then its argument prototype would still apply.  */
| char gettimeofday ();
| /* The GNU C library defines this for functions which it implements
|     to always fail with ENOSYS.  Some functions are actually named
|     something starting with __ and the normal name is an alias.  */
| #if defined (__stub_gettimeofday) || defined (__stub___gettimeofday)
| choke me
| #else
| char (*f) () = gettimeofday;
| #endif
| #ifdef __cplusplus
| }
| #endif
| 
| int
| main ()
| {
| return f != gettimeofday;
|   ;
|   return 0;
| }
configure:9204: result: no
configure:9134: checking for select
configure:9184: cc -o conftest  -O   conftest.c -lnsl -lcurses -lHcurses -lm  -lc >&5
cc: "configure", line 9234: error 1584: Inconsistent type declaration: "select".
cc: "configure", line 9241: warning 604: Pointers are not assignment-compatible.
cc: "configure", line 9250: warning 605: Illegal pointer combination for !=.
configure:9187: $? = 1
configure: failed program was:
| #line 9139 "configure"
| /* confdefs.h.  */
| 
| #define PACKAGE_NAME ""
| #define PACKAGE_TARNAME ""
| #define PACKAGE_VERSION ""
| #define PACKAGE_STRING ""
| #define PACKAGE_BUGREPORT ""
| #define MACHTYPE "hppa2.0w"
| #define VENDOR "hp"
| #define OSTYPE "hpux11.11"
| #define GLOBAL_ZSHENV "/etc/zshenv"
| #define GLOBAL_ZSHRC "/etc/zshrc"
| #define GLOBAL_ZPROFILE "/etc/zprofile"
| #define GLOBAL_ZLOGIN "/etc/zlogin"
| #define GLOBAL_ZLOGOUT "/etc/zlogout"
| #define RESTRICTED_R 1
| #define CONFIG_LOCALE 1
| #define PROTOTYPES 1
| #define HAVE_ALLOCA_H 1
| #define HAVE_ALLOCA 1
| #define HAVE_UNION_INIT 1
| #define HAVE_VARIABLE_LENGTH_ARRAYS 1
| #define HAVE_DIRENT_H 1
| #define STDC_HEADERS 1
| #define TIME_WITH_SYS_TIME 1
| #define HAVE_SYS_WAIT_H 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_SYS_STAT_H 1
| #define HAVE_STDLIB_H 1
| #define HAVE_STRING_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_STRINGS_H 1
| #define HAVE_INTTYPES_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_SYS_TIME_H 1
| #define HAVE_SYS_TIMES_H 1
| #define HAVE_TERMIO_H 1
| #define HAVE_TERMIOS_H 1
| #define HAVE_SYS_PARAM_H 1
| #define HAVE_STRING_H 1
| #define HAVE_MEMORY_H 1
| #define HAVE_LIMITS_H 1
| #define HAVE_FCNTL_H 1
| #define HAVE_SYS_UTSNAME_H 1
| #define HAVE_SYS_RESOURCE_H 1
| #define HAVE_LOCALE_H 1
| #define HAVE_ERRNO_H 1
| #define HAVE_STDIO_H 1
| #define HAVE_STDLIB_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_UTMP_H 1
| #define HAVE_UTMPX_H 1
| #define HAVE_SYS_TYPES_H 1
| #define HAVE_PWD_H 1
| #define HAVE_GRP_H 1
| #define HAVE_POLL_H 1
| #define HAVE_SYS_MMAN_H 1
| #define HAVE_NETINET_IN_SYSTM_H 1
| #define HAVE_LANGINFO_H 1
| #define HAVE_WCHAR_H 1
| #define HAVE_STDDEF_H 1
| #define HAVE_SYS_STROPTS_H 1
| #define HAVE_DLFCN_H 1
| #define HAVE_DL_H 1
| #define HAVE_LIBM 1
| #define HAVE_CURSES_H 1
| #define HAVE_TERM_H 1
| #define RETSIGTYPE void
| #define ZSH_64_BIT_TYPE long long
| #define ZSH_64_BIT_UTYPE unsigned long long
| #define HAVE_STRUCT_TIMEZONE 1
| #define HAVE_STRUCT_UTMP 1
| #define HAVE_STRUCT_UTMPX 1
| #define HAVE_STRUCT_UTMP_UT_HOST 1
| #define HAVE_STRUCT_UTMPX_UT_HOST 1
| #define HAVE_STRUCT_UTMPX_UT_TV 1
| #define HAVE_STRUCT_DIRENT_D_INO 1
| #define HAVE_STRUCT_SOCKADDR_IN6_SIN6_SCOPE_ID 1
| /* end confdefs.h.  */
| /* System header to define __stub macros and hopefully few prototypes,
|     which can conflict with char select (); below.
|     Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
|     <limits.h> exists even on freestanding compilers.  */
| #ifdef __STDC__
| # include <limits.h>
| #else
| # include <assert.h>
| #endif
| /* Override any gcc2 internal prototype to avoid an error.  */
| #ifdef __cplusplus
| extern "C"
| {
| #endif
| /* We use char because int might match the return type of a gcc2
|    builtin and then its argument prototype would still apply.  */
| char select ();
| /* The GNU C library defines this for functions which it implements
|     to always fail with ENOSYS.  Some functions are actually named
|     something starting with __ and the normal name is an alias.  */
| #if defined (__stub_select) || defined (__stub___select)
| choke me
| #else
| char (*f) () = select;
| #endif
| #ifdef __cplusplus
| }
| #endif
| 
| int
| main ()
| {
| return f != select;
|   ;
|   return 0;
| }
configure:9204: result: no

Re: 4.2.0-pre-1 on HP-UX 11.11

[Peter Stephenson]

Paul Ackersviller wrote:
> I've attached the portion in question of my config.log -- I don't
> understand why it appears to be trying signatures of char (*)().

Thanks.  Hmm, I think autoconf 2.59 may have fixed this... I'll make
sure I generate -pre-3 with that.  -pre-2 uses 2.57.

(My Linux box at work as 2.57 and it's slightly more convenient to
generate releases there than on the Solaris system which has
2.59... usual story.  I'll upgrade.)

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************