naviserver-devel

[naviserver-devel] compilation ?

[Zoran V. <zv...@ar...>]

Hi!

I'm afraid the compilation is somehow broken:

zoran:~/sf/naviserver/nsd zoran$ make
gcc -pipe -g -Wall -Wno-implicit-int -fno-common  -I../include -I"/ 
usr/local/include"  -DNO_CONST -DHAVE_CONFIG_H   -c -o log.o log.c
log.c: In function `LogTime':
log.c:762: error: wrong type argument to unary minus
log.c: At top level:
log.c:37: warning: `RCSID' defined but not used
make: *** [log.o] Error 1

I have re-run the configure with --enable-symbols --enable-threads
but I somehow miss many of the usual CFLAGS normally defined when
I do the compilation. This is normally what I'd expected to see:

gcc -pipe -DURLDECODE_RELAXED=1 -g -Wall -Wno-implicit-int -fno- 
strict-aliasing -fPIC  -I../include -I/usr/local/include  -DNO_CONST   
-DTCL_THREADS=1 -DUSE_THREAD_ALLOC=1 -D_REENTRANT=1 -D_THREAD_SAFE=1 - 
DHAVE_PTHREAD_ATTR_SETSTACKSIZE=1 -DHAVE_PTHREAD_ATFORK=1 - 
DHAVE_READDIR_R=1 -DHAVE_THREE_ARG_READDIR_R=1 -DPEEK_XCLOSEIM=1 - 
D_LARGEFILE64_SOURCE=1 -DTCL_WIDE_INT_TYPE=long\ long - 
DHAVE_STRUCT_STAT64=1 -DHAVE_OPEN64=1 -DHAVE_LSEEK64=1 - 
DHAVE_TYPE_OFF64_T=1 -DHAVE_GETCWD=1 -DHAVE_OPENDIR=1 -DHAVE_STRSTR=1  
-DHAVE_STRTOL=1 -DHAVE_STRTOLL=1 -DHAVE_STRTOULL=1 -DHAVE_TMPNAM=1 - 
DHAVE_WAITPID=1 -DHAVE_LIMITS_H=1 -DHAVE_UNISTD_H=1 - 
DHAVE_SYS_PARAM_H=1 -DUSE_TERMIOS=1 -DHAVE_SYS_TIME_H=1 - 
DTIME_WITH_SYS_TIME=1 -DHAVE_TM_ZONE=1 -DHAVE_GMTIME_R=1 - 
DHAVE_LOCALTIME_R=1 -DHAVE_TM_GMTOFF=1 -DHAVE_TIMEZONE_VAR=1 - 
DHAVE_ST_BLKSIZE=1 -DSTDC_HEADERS=1 -DHAVE_SIGNED_CHAR=1 - 
DHAVE_LANGINFO=1 -DHAVE_SYS_IOCTL_H=1   -DTCL_CFG_DEBUG=1 - 
DHAVE_INTTYPES_H=1 -DHAVE_TIMEGM=1 -DHAVE_POLL=1 -DHAVE_DRAND48=1 - 
DHAVE_RANDOM=1 -DHAVE_CMMSG=1 -DHAVE_GETADDRINFO=1 - 
DHAVE_GETNAMEINFO=1 -DHAVE_ZLIB_H=1 -DHAVE_LIBZ=1  -DNsdInit=_init   - 
c -o set.o set.c

Where have all those defines gone?

Zoran

Re: [naviserver-devel] compilation ?

[Zoran V. <zv...@ar...>]

Am 09.06.2005 um 22:49 schrieb Zoran Vasiljevic:

> Hi!
>
> I'm afraid the compilation is somehow broken:
>
> zoran:~/sf/naviserver/nsd zoran$ make
> gcc -pipe -g -Wall -Wno-implicit-int -fno-common  -I../include -I"/ 
> usr/local/include"  -DNO_CONST -DHAVE_CONFIG_H   -c -o log.o log.c
> log.c: In function `LogTime':
> log.c:762: error: wrong type argument to unary minus
> log.c: At top level:
> log.c:37: warning: `RCSID' defined but not used
> make: *** [log.o] Error 1


I see... the config.h is now used. But I'd need some help
to know how this all is used. Any hints for startup?

OTOH, the HAVE_TM_GMTOFF define is missing. W/o this the
Darwin compilation fails as seen from above.

Zoran

Re: [naviserver-devel] compilation ?

[Vlad S. <vl...@cr...>]

I just fixed that in the CVS

Zoran Vasiljevic wrote:
> Hi!
> 
> I'm afraid the compilation is somehow broken:
> 
> zoran:~/sf/naviserver/nsd zoran$ make
> gcc -pipe -g -Wall -Wno-implicit-int -fno-common  -I../include -I"/ 
> usr/local/include"  -DNO_CONST -DHAVE_CONFIG_H   -c -o log.o log.c
> log.c: In function `LogTime':
> log.c:762: error: wrong type argument to unary minus
> log.c: At top level:
> log.c:37: warning: `RCSID' defined but not used
> make: *** [log.o] Error 1
> 
> I have re-run the configure with --enable-symbols --enable-threads
> but I somehow miss many of the usual CFLAGS normally defined when
> I do the compilation. This is normally what I'd expected to see:
> 
> gcc -pipe -DURLDECODE_RELAXED=1 -g -Wall -Wno-implicit-int -fno- 
> strict-aliasing -fPIC  -I../include -I/usr/local/include  -DNO_CONST   
> -DTCL_THREADS=1 -DUSE_THREAD_ALLOC=1 -D_REENTRANT=1 -D_THREAD_SAFE=1 - 
> DHAVE_PTHREAD_ATTR_SETSTACKSIZE=1 -DHAVE_PTHREAD_ATFORK=1 - 
> DHAVE_READDIR_R=1 -DHAVE_THREE_ARG_READDIR_R=1 -DPEEK_XCLOSEIM=1 - 
> D_LARGEFILE64_SOURCE=1 -DTCL_WIDE_INT_TYPE=long\ long - 
> DHAVE_STRUCT_STAT64=1 -DHAVE_OPEN64=1 -DHAVE_LSEEK64=1 - 
> DHAVE_TYPE_OFF64_T=1 -DHAVE_GETCWD=1 -DHAVE_OPENDIR=1 -DHAVE_STRSTR=1  
> -DHAVE_STRTOL=1 -DHAVE_STRTOLL=1 -DHAVE_STRTOULL=1 -DHAVE_TMPNAM=1 - 
> DHAVE_WAITPID=1 -DHAVE_LIMITS_H=1 -DHAVE_UNISTD_H=1 - 
> DHAVE_SYS_PARAM_H=1 -DUSE_TERMIOS=1 -DHAVE_SYS_TIME_H=1 - 
> DTIME_WITH_SYS_TIME=1 -DHAVE_TM_ZONE=1 -DHAVE_GMTIME_R=1 - 
> DHAVE_LOCALTIME_R=1 -DHAVE_TM_GMTOFF=1 -DHAVE_TIMEZONE_VAR=1 - 
> DHAVE_ST_BLKSIZE=1 -DSTDC_HEADERS=1 -DHAVE_SIGNED_CHAR=1 - 
> DHAVE_LANGINFO=1 -DHAVE_SYS_IOCTL_H=1   -DTCL_CFG_DEBUG=1 - 
> DHAVE_INTTYPES_H=1 -DHAVE_TIMEGM=1 -DHAVE_POLL=1 -DHAVE_DRAND48=1 - 
> DHAVE_RANDOM=1 -DHAVE_CMMSG=1 -DHAVE_GETADDRINFO=1 - DHAVE_GETNAMEINFO=1 
> -DHAVE_ZLIB_H=1 -DHAVE_LIBZ=1  -DNsdInit=_init   - c -o set.o set.c
> 
> Where have all those defines gone?
> 
> Zoran
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: NEC IT Guy Games.  How far can you 
> shotput
> a projector? How fast can you ride your desk chair down the office luge 
> track?
> If you want to score the big prize, get to know the little guy.  Play to 
> win an NEC 61" plasma display: http://www.necitguy.com/?r=20
> _______________________________________________
> naviserver-devel mailing list
> nav...@li...
> https://lists.sourceforge.net/lists/listinfo/naviserver-devel

-- 
Vlad Seryakov
571 262-8608 office
vl...@cr...
http://www.crystalballinc.com/vlad/

Re: [naviserver-devel] compilation ?

[Zoran V. <zv...@ar...>]

Am 09.06.2005 um 23:05 schrieb Vlad Seryakov:

> I just fixed that in the CVS

Hmmm... now it does compile but I still do not understand one thing:
include/config.h file overlapps with TCL_DEFS in various places.
Is this willingly done so or is this just a coincidence?

Also, do we need TCL_DEFS when using the config.h?
Stephen has done some work on that lately as I see from
the ChangeLog. Stephen, can you mabe comment on this?

Cheers
Zoran

Re: [naviserver-devel] compilation ?

[Stephen D. <sd...@gm...>]

On 6/9/05, Zoran Vasiljevic <zv...@ar...> wrote:
>=20
> Am 09.06.2005 um 23:05 schrieb Vlad Seryakov:
>=20
> > I just fixed that in the CVS
>
> Hmmm... now it does compile but I still do not understand one thing:
> include/config.h file overlapps with TCL_DEFS in various places.
> Is this willingly done so or is this just a coincidence?
>=20
> Also, do we need TCL_DEFS when using the config.h?
> Stephen has done some work on that lately as I see from
> the ChangeLog. Stephen, can you mabe comment on this?


Yes, config.h is the new TCL_DEFS -- adding TCL_DEFS back into
Makefile.global.in is not the right fix.

Zoran, is HAVE_TM_GMTOFF defined in config.h after you run configure?=20
The check in tcl.m4 which we now ship looks identical to the code in
the Tcl tarball.

Re: [naviserver-devel] compilation ?

[Vlad S. <vl...@cr...>]

HAVE_TM_GMTOFF is not defined in the config.h.in
so it never gets to config.h

The problem i usually have with config.h if i include different packages
and they all use config.h which is common with GNU/configure based 
packages they interfere with each other and config.h should be exported 
and distributed with public header files. It was designed for internal 
compilation only(this is my guess), now even public API cannot be 
compiled without it.

I think old style having TCL_DEFs in the Makefile was more cleaner way.

Stephen Deasey wrote:
> On 6/9/05, Zoran Vasiljevic <zv...@ar...> wrote:
> 
>>Am 09.06.2005 um 23:05 schrieb Vlad Seryakov:
>>
>>
>>>I just fixed that in the CVS
>>
>>Hmmm... now it does compile but I still do not understand one thing:
>>include/config.h file overlapps with TCL_DEFS in various places.
>>Is this willingly done so or is this just a coincidence?
>>
>>Also, do we need TCL_DEFS when using the config.h?
>>Stephen has done some work on that lately as I see from
>>the ChangeLog. Stephen, can you mabe comment on this?
> 
> 
> 
> Yes, config.h is the new TCL_DEFS -- adding TCL_DEFS back into
> Makefile.global.in is not the right fix.
> 
> Zoran, is HAVE_TM_GMTOFF defined in config.h after you run configure? 
> The check in tcl.m4 which we now ship looks identical to the code in
> the Tcl tarball.
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: NEC IT Guy Games.  How far can you shotput
> a projector? How fast can you ride your desk chair down the office luge track?
> If you want to score the big prize, get to know the little guy.  
> Play to win an NEC 61" plasma display: http://www.necitguy.com/?r 
> _______________________________________________
> naviserver-devel mailing list
> nav...@li...
> https://lists.sourceforge.net/lists/listinfo/naviserver-devel

-- 
Vlad Seryakov
571 262-8608 office
vl...@cr...
http://www.crystalballinc.com/vlad/

Re: [naviserver-devel] compilation ?

[Stephen D. <sd...@gm...>]

On 6/9/05, Vlad Seryakov <vl...@cr...> wrote:
> HAVE_TM_GMTOFF is not defined in the config.h.in
> so it never gets to config.h


Ah, thanks.  I've added the right call in the configure script to make
this happen.


> The problem i usually have with config.h if i include different packages
> and they all use config.h which is common with GNU/configure based
> packages they interfere with each other and config.h should be exported
> and distributed with public header files. It was designed for internal
> compilation only(this is my guess), now even public API cannot be
> compiled without it.


When compiling a naviserver module the Tcl defs will be present either
on the command line because they're embeded in Makefile.global or in
the config.h header, so I don't think it makes a difference from this
point of view.

Re: [naviserver-devel] compilation ?

[Zoran V. <zv...@ar...>]

Am 10.06.2005 um 08:21 schrieb Stephen Deasey:

> On 6/9/05, Vlad Seryakov <vl...@cr...> wrote:
>
>> HAVE_TM_GMTOFF is not defined in the config.h.in
>> so it never gets to config.h
>>
>
>
> Ah, thanks.  I've added the right call in the configure script to make
> this happen.

Still some things left:

dsprintf.c: In function `Ns_DStringVPrintf':
dsprintf.c:294: warning: use of `long double' type; its size may  
change in a future release
dsprintf.c:294: warning: (Long double usage is reported only once for  
each file.
dsprintf.c:294: warning: To disable this warning, use -Wno-long-double.)

What should we do about that?


return.c: In function `Ns_ConnConstructHeaders':
return.c:262: warning: suggest parentheses around && within ||
return.c: At top level:

This is here for the long time. I think it is the time to
get rid of it. I looked there but am not sure.


tclxkeylist.c: In function `Tcl_GetKeyedListKeys':
tclxkeylist.c:285: warning: passing arg 2 of `Tcl_GetStringFromObj'  
from incompatible pointer type
tclxkeylist.c:293: warning: passing arg 2 of `Tcl_GetStringFromObj'  
from incompatible pointer type
tclxkeylist.c: In function `Tcl_GetKeyedListField':
tclxkeylist.c:353: warning: passing arg 2 of `Tcl_GetStringFromObj'  
from incompatible pointer type
tclxkeylist.c: In function `Tcl_SetKeyedListField':
tclxkeylist.c:407: warning: passing arg 2 of `Tcl_GetStringFromObj'  
from incompatible pointer type
tclxkeylist.c: In function `Tcl_DeleteKeyedListField':
tclxkeylist.c:453: warning: passing arg 2 of `Tcl_GetStringFromObj'  
from incompatible pointer type
tclxkeylist.c: At top level:

This should be easy. I will take care of that.

Stepen, you mentioned usage of some macro to get rid of the

nsdb.c:36: warning: `RCSID' defined but not used

type of warnings. I suppose you'd have to go wholesale change all  
files, right?

Anyways, it compiles now on Darwin. Not very clean though:

ld: warning prebinding disabled because (__TEXT segment (address =  
0x0 size = 0x63000) of libnsd.dylib overlaps with __TEXT segment  
(address = 0x0 size = 0x6000) of /usr/local/ns/lib/libnsthread.dylib
ld: warning prebinding disabled because (__TEXT segment (address =  
0x0 size = 0x63000) of libnsd.dylib overlaps with __DATA segment  
(address = 0x6000 size = 0x1000) of /usr/local/ns/lib/libnsthread.dylib
ld: warning prebinding disabled because (__TEXT segment (address =  
0x0 size = 0x63000) of libnsd.dylib overlaps with __LINKEDIT segment  
(address = 0x7000 size = 0xd000) of /usr/local/ns/lib/libnsthread.dylib

but I will have to see why. The tcl.m4 has the "-prebind" option to  
linker
and I'd have to check if this is really needed.

Zoran

Re: [naviserver-devel] compilation ?

[Stephen D. <sd...@gm...>]

On 6/10/05, Zoran Vasiljevic <zv...@ar...> wrote:
>=20
> Am 10.06.2005 um 08:21 schrieb Stephen Deasey:
>=20
> > On 6/9/05, Vlad Seryakov <vl...@cr...> wrote:
> >
> >> HAVE_TM_GMTOFF is not defined in the config.h.in
> >> so it never gets to config.h
> >>
> >
> >
> > Ah, thanks.  I've added the right call in the configure script to make
> > this happen.
>=20
> Still some things left:
>=20
> dsprintf.c: In function `Ns_DStringVPrintf':
> dsprintf.c:294: warning: use of `long double' type; its size may
> change in a future release
> dsprintf.c:294: warning: (Long double usage is reported only once for
> each file.
> dsprintf.c:294: warning: To disable this warning, use -Wno-long-double.)
>=20
> What should we do about that?


Has this always been a warning on Darwin?  I don't see any LONGDBL
mentioned in the Tcl source...


> return.c: In function `Ns_ConnConstructHeaders':
> return.c:262: warning: suggest parentheses around && within ||
> return.c: At top level:
>=20
> This is here for the long time. I think it is the time to
> get rid of it. I looked there but am not sure.


I cleaned up some warnings a while ago but left this one as there's a
better solution in AOLserver HEAD, but it requires a few extra new
functions which I didn't want to add just for a clean-up.


> tclxkeylist.c: In function `Tcl_GetKeyedListKeys':
> tclxkeylist.c:285: warning: passing arg 2 of `Tcl_GetStringFromObj'
> from incompatible pointer type
> tclxkeylist.c:293: warning: passing arg 2 of `Tcl_GetStringFromObj'
> from incompatible pointer type
> tclxkeylist.c: In function `Tcl_GetKeyedListField':
> tclxkeylist.c:353: warning: passing arg 2 of `Tcl_GetStringFromObj'
> from incompatible pointer type
> tclxkeylist.c: In function `Tcl_SetKeyedListField':
> tclxkeylist.c:407: warning: passing arg 2 of `Tcl_GetStringFromObj'
> from incompatible pointer type
> tclxkeylist.c: In function `Tcl_DeleteKeyedListField':
> tclxkeylist.c:453: warning: passing arg 2 of `Tcl_GetStringFromObj'
> from incompatible pointer type
> tclxkeylist.c: At top level:
>=20
> This should be easy. I will take care of that.
>=20
> Stepen, you mentioned usage of some macro to get rid of the
>=20
> nsdb.c:36: warning: `RCSID' defined but not used
>=20
> type of warnings. I suppose you'd have to go wholesale change all
> files, right?


Yes, that's the RFE I posted recently.  I wasn't going to add that
until after this release, but we've started churning the code again so
might as well add this now I suppose.

As far as I remember, adding some of the extra checks such as for
printf format strings turns up some genuine bugs.  Maybe we should go
for those now too?


> Anyways, it compiles now on Darwin. Not very clean though:
>=20
> ld: warning prebinding disabled because (__TEXT segment (address =3D
> 0x0 size =3D 0x63000) of libnsd.dylib overlaps with __TEXT segment
> (address =3D 0x0 size =3D 0x6000) of /usr/local/ns/lib/libnsthread.dylib
> ld: warning prebinding disabled because (__TEXT segment (address =3D
> 0x0 size =3D 0x63000) of libnsd.dylib overlaps with __DATA segment
> (address =3D 0x6000 size =3D 0x1000) of /usr/local/ns/lib/libnsthread.dyl=
ib
> ld: warning prebinding disabled because (__TEXT segment (address =3D
> 0x0 size =3D 0x63000) of libnsd.dylib overlaps with __LINKEDIT segment
> (address =3D 0x7000 size =3D 0xd000) of /usr/local/ns/lib/libnsthread.dyl=
ib
>=20
> but I will have to see why. The tcl.m4 has the "-prebind" option to
> linker
> and I'd have to check if this is really needed.


If this is a bug you should also file it with the Tcl project as this
is the tcl.m4 used by all Tcl packages.

Re: [naviserver-devel] compilation ?

[Zoran V. <zv...@ar...>]

Am 10.06.2005 um 09:09 schrieb Stephen Deasey:
>>
>> dsprintf.c: In function `Ns_DStringVPrintf':
>> dsprintf.c:294: warning: use of `long double' type; its size may
>> change in a future release
>> dsprintf.c:294: warning: (Long double usage is reported only once for
>> each file.
>> dsprintf.c:294: warning: To disable this warning, use -Wno-long- 
>> double.)
>>
>> What should we do about that?
>>
>
>
> Has this always been a warning on Darwin?  I don't see any LONGDBL
> mentioned in the Tcl source...

Aha! No. It actually popped out since I got 10.4 installed.
I will see what is going on here.

>
>
>
>> return.c: In function `Ns_ConnConstructHeaders':
>> return.c:262: warning: suggest parentheses around && within ||
>> return.c: At top level:
>>
>> This is here for the long time. I think it is the time to
>> get rid of it. I looked there but am not sure.
>>
>
>
> I cleaned up some warnings a while ago but left this one as there's a
> better solution in AOLserver HEAD, but it requires a few extra new
> functions which I didn't want to add just for a clean-up.
>

OK.

>
> As far as I remember, adding some of the extra checks such as for
> printf format strings turns up some genuine bugs.  Maybe we should go
> for those now too?

Certainly.

>
> If this is a bug you should also file it with the Tcl project as this
> is the tcl.m4 used by all Tcl packages.
>

I know. But this is not a bug per-se. It is somehow related to the
linking we're doing. Again, I will look into this.

Zoran