Very Computer

Has anyone compiled mutt 1.2.5 on hp-ux 10.20? If so, how? I'm getting
errors.

Thanks,
Ljubomir Josifovski

I've compiled mutt 1.2i on HP-UX 10.20.  Have you compiled earlier
versions of mutt on HP-UX without problems?  What errors are you
getting?

Gary

I have had prblems with mutt 1.2.5 on HPUX as far as I recall 1.0i was fine (and I am still using it as the < key fails to work in 1.2.5 which is very annoying

Some clues below I am using HP ansi compiler and gnu make

        Jeff

With ncurses 4.2
# ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses

cc -Aa -D_HPUX_SOURCE -DSHAREDIR=\"/opt/mutt-1.2.5/lib/mutt\" -DSYSCONFDIR=\"/o
pt/mutt-1.2.5/etc\"      -DBINDIR=\"/opt/mutt-1.2.5/bin\" -DHAVE_CONFIG_H=1 -I.
 -I.  -Iintl -I/opt/mutt-1.2.5/include  -I./intl  -g -c keymap.c
cc: "keymap.c", line 69: error 1588: "KEY_END" undefined.
cc: "keymap.c", line 69: error 1521: Incorrect initialization.
gmake[1]: *** [keymap.o] Error 1
gmake[1]: Leaving directory `/usr/source/mutt-1.2.5'
gmake: *** [install-recursive] Error 1

with slang
# ./configure --prefix=/opt/mutt-1.2.5 --with-slang=/opt/slang

compiles but no colour

with HP curses

# ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/usr/lib

cc -Aa -D_HPUX_SOURCE -DSHAREDIR=\"/opt/mutt-1.2.5/lib/mutt\" -DSYSCONFDIR=\"/o
pt/mutt-1.2.5/etc\"      -DBINDIR=\"/opt/mutt-1.2.5/bin\" -DHAVE_CONFIG_H=1 -I.
 -I.  -Iintl -I/opt/mutt-1.2.5/include  -I/usr/lib/include -I./intl  -g -c keym
ap.c
cc: "keymap.c", line 69: error 1588: "KEY_END" undefined.
cc: "keymap.c", line 69: error 1521: Incorrect initialization.
gmake[1]: *** [keymap.o] Error 1
gmake[1]: Leaving directory `/usr/source/mutt-1.2.5'
gmake: *** [install-recursive] Error 1
#

needs color explicitly
# ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses --enable-colo
r=YES
try
# rm ./config.cache before changes

now have color but < key acts wierdly! (same in slang but no colour)

 > I have had prblems with mutt 1.2.5 on HPUX

What terminal emulator and $TERM do you use? Have tried alternatives?
And likely the curses you actually took was broken:

 > With ncurses 4.2
 > # ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses

 > cc -Aa -D_HPUX_SOURCE -DSHAREDIR=\"/opt/mutt-1.2.5/lib/mutt\"
 > -DSYSCONFDIR=\"/opt/mutt-1.2.5/etc\"  -DBINDIR=\"/opt/mutt-1.2.5/bin\"
 > -DHAVE_CONFIG_H=1 -I.  -I.  -Iintl -I/opt/mutt-1.2.5/include  -I./intl
 >  -g -c keymap.c
 > cc: "keymap.c", line 69: error 1588: "KEY_END" undefined.

Your system curses.h was taken, not ncurses-4.2's.
(/usr/include/curses.h doesn't define KEY_END)

Apparently your configure went wrong, as there's
no "-I/opt/ncurses/include" above.

With both ncurses-5.0 and ncurses-4.2 it's all fine here.
(gcc-2.95.2 (no ANSI cc on board), GNU make).

BTW, there's also a system curses with color on HP-UX.
/usr/include/curses_colr/curses.h (i.e.  <curses_colr/curses.h>)
and /usr/lib/libcur_colr.*   But i never got happy with it.

Sven

F'up to myself...

But before: i just want to point out again, that i
have not meant *ncurses-4.2* being broken:

 > And likely the curses you actually took was broken:

 >  > With ncurses 4.2
 >  > # ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses

  [but it turns out that another curses.h was taken]

Now i have tried with the *system curses* (HPUX10.20) also
and fiddled out that the mutt source uses macros which
are not included in the section in <curses.h> which is actually taken:

[...]
#if !defined(_XOPEN_SOURCE_EXTENDED) || defined(__HP_CURSES)
   <section which is taken>
#else
   <needed macros>
#fi
[...]

By (globally) defining '_XOPEN_SOURCE_EXTENDED' and
'_ACS_COMPAT_CODE' (which is later on needed also),
i got it compile without additional warnings.

I haven't tested it thoroughly, but with just colors commented out
from my .muttrc i couldn't find any problems when mailing a bit
with myself (using the internal pager).

I usually don't use HP-UX and also i have no clue what exactly
the consequences of these variables mean.  Thus this might be
the wrong approach.

I also tried the system-curses-with-colors again
('<curses_colr/curses.h>' and '-lcur_colr', adjusting the Makefile)
this time with the above macros, but i couldn't get color working
for the 1st try and stopped there (as i guess one really should
get ncurses/slang for color anyway, instead of hacking here) .

Sven

F'up to myself...

But before: i just want to point out again, that i
have not meant *ncurses-4.2* being broken:

 > And likely the curses you actually took was broken:

 >  > With ncurses 4.2
 >  > # ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses

  [but it turns out that another curses.h was taken]

[ supersede: the following is only for gcc-2.95.2, GNU make ]

Now i have tried with the *system curses* (HPUX10.20) also
and fiddled out that the mutt source uses macros which
are not included in the section in <curses.h> which is actually taken:

[...]
#if !defined(_XOPEN_SOURCE_EXTENDED) || defined(__HP_CURSES)
   <section which is taken>
#else
   <needed macros>
#fi
[...]

By (globally) defining '_XOPEN_SOURCE_EXTENDED' and
'_ACS_COMPAT_CODE' (which is later on needed also),
i got it compile without additional warnings.

I haven't tested it thoroughly, but with just colors commented out
from my .muttrc i couldn't find any problems when mailing a bit
with myself (using the internal pager).

I usually don't use HP-UX and also i have no clue what exactly
the consequences of these variables mean.  Thus this might be
the wrong approach.

I also tried the system-curses-with-colors again
('<curses_colr/curses.h>' and '-lcur_colr', adjusting the Makefile)
this time with the above macros, but i couldn't get color working
for the 1st try and stopped there (as i guess one really should
get ncurses/slang for color anyway, instead of hacking here) .

Sven

I am using dtterm and $TERM=dtterm, also works fine on putty from a windows pc. HP supplied xterm gives no colour

Quote:

>  > With ncurses 4.2
>  > # ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses

>  > cc -Aa -D_HPUX_SOURCE -DSHAREDIR=\"/opt/mutt-1.2.5/lib/mutt\"
>  > -DSYSCONFDIR=\"/opt/mutt-1.2.5/etc\"  -DBINDIR=\"/opt/mutt-1.2.5/bin\"
>  > -DHAVE_CONFIG_H=1 -I.  -I.  -Iintl -I/opt/mutt-1.2.5/include  -I./intl
>  >  -g -c keymap.c
>  > cc: "keymap.c", line 69: error 1588: "KEY_END" undefined.

> Your system curses.h was taken, not ncurses-4.2's.
> (/usr/include/curses.h doesn't define KEY_END)

> Apparently your configure went wrong, as there's
> no "-I/opt/ncurses/include" above.

Thanks very much. It must have been the earlier attempts. I went back to the
distribution and used
 ./configure --prefix=/opt/mutt-1.2.5 --with-curses=/opt/ncurses --enable-color=YES

then all was okay except for a pesky problem with the '<' key which
needs to be pressed twice befor it produces a character on screen. It
is okay in 1.0i. Any other reports of this? is it a HP thing? The '>'
key is fine as are all others as far as I have seen.

        Jeff

Quote:

> With both ncurses-5.0 and ncurses-4.2 it's all fine here.
> (gcc-2.95.2 (no ANSI cc on board), GNU make).

> BTW, there's also a system curses with color on HP-UX.
> /usr/include/curses_colr/curses.h (i.e.  <curses_colr/curses.h>)
> and /usr/lib/libcur_colr.*   But i never got happy with it.

> Sven