Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: exim: dev

Candidate for 4.52 release

 

 

exim dev RSS feed   Index | Next | Previous | View Threaded


ph10 at cus

Jun 23, 2005, 6:59 AM

Post #1 of 5 (166 views)
Permalink
Candidate for 4.52 release

There's a snapshot in

ftp://ftp.csx.cam.ac.uk/pub/software/email/exim/Testing/exim-snapshot.tar.gz
ftp://ftp.csx.cam.ac.uk/pub/software/email/exim/Testing/exim-snapshot.tar.gz.sig

which is the first candidate for a 4.52 release. Please test. I intend
to release it sometime next week.

--
Philip Hazel University of Cambridge Computing Service,
ph10 [at] cus Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book


mb at dcs

Jun 23, 2005, 9:34 AM

Post #2 of 5 (157 views)
Permalink
Re: Candidate for 4.52 release [In reply to]

Philip Hazel wrote:

> ftp://ftp.csx.cam.ac.uk/pub/software/email/exim/Testing/exim-snapshot.tar.gz
>
> which is the first candidate for a 4.52 release. Please test.

In "scripts/os-type", the following code is slightly broken (please
excuse the sloppy cut-n-paste):

# In the case of Linux we need to distinguish which libc is used.
# This is more cautious than it needs to be. In practice libc5 will always
# be a symlink, and libc6 will always be a linker control file, but it's
# easy enough to do a better check, and check the symlink destination or the
# control file contents and make sure.

Linux) if [ -L /usr/lib/libc.so ]; then
if [ x"$(file /usr/lib/libc.so | grep "libc.so.5")"x != xx
]; then
os=Linux-libc5
fi
else
if grep -q libc.so.5 /usr/lib/libc.so; then
os=Linux-libc5
fi
fi
;;

The "grep -q" spits annoyingly when libc.so lives in lib64, like on
x86_64 Fedora Core. The "if" fails, mercifully :)

Rather than fix it, perhaps "Linux-libc5" might like to be dropped as a
build? Does anyone still use it?

Slightly controversially,

Matt


OpenMacNews at speakeasy

Jun 23, 2005, 4:49 PM

Post #3 of 5 (157 views)
Permalink
Re: Candidate for 4.52 release -- make failure, "No rule to make target `lookups/lookups.a', needed by `exim'" [In reply to]

hi all,

building 4.52rc on OSX 10.4.1, a prior issue w/ libspf2 api has been cured.

tho, next 'make' failure is:

...

cc exim_monitor/em_version.c
cc -o eximon.bin
/usr/bin/ld: warning multiple definitions of symbol __XawTextSearch
em_TextPop.o definition of __XawTextSearch in section (__TEXT,__text)
/usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of __XawTextSearch
/usr/bin/ld: warning multiple definitions of symbol __XawTextDoSearchAction
em_TextPop.o definition of __XawTextDoSearchAction in section
(__TEXT,__text)
/usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of __XawTextDoSearchAction
/usr/bin/ld: warning multiple definitions of symbol
__XawTextPopdownSearchAction
em_TextPop.o definition of __XawTextPopdownSearchAction in section
(__TEXT,__text)
/usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of
__XawTextPopdownSearchAction
>>> exim monitor binary built

cc dnsdb.c
dnsdb.c:45: error: 'T_A' undeclared here (not in a function)
dnsdb.c:52: error: 'T_CNAME' undeclared here (not in a function)
dnsdb.c:54: error: 'T_MX' undeclared here (not in a function)
dnsdb.c:56: error: 'T_NS' undeclared here (not in a function)
dnsdb.c:57: error: 'T_PTR' undeclared here (not in a function)
dnsdb.c: In function 'dnsdb_find':
dnsdb.c:223: warning: comparison between pointer and integer
dnsdb.c:235: warning: pointer/integer type mismatch in conditional
expression
dnsdb.c:235: warning: pointer/integer type mismatch in conditional
expression
dnsdb.c:235: warning: initialization makes integer from pointer without a
cast
dnsdb.c:243: warning: comparison between pointer and integer
dnsdb.c:283: warning: comparison between pointer and integer
dnsdb.c:320: warning: comparison between pointer and integer
make[2]: *** [dnsdb.o] Error 1

make[2]: `routers.a' is up to date.

make[2]: `transports.a' is up to date.

make[2]: `auths.a' is up to date.

make[1]: *** No rule to make target `lookups/lookups.a', needed by `exim'.
Stop.
make: *** [go] Error 2


fwiw, my '/usr/X11R6/lib/libXaw.dylib' is from a build of X.Org's X11R6
v6.8.99.11.

now to sleuth ... lest any advice?

richard


ph10 at cus

Jun 24, 2005, 3:19 AM

Post #4 of 5 (157 views)
Permalink
Re: Candidate for 4.52 release [In reply to]

On Thu, 23 Jun 2005, Matt Bernstein wrote:

> Rather than fix it, perhaps "Linux-libc5" might like to be dropped as a build?
> Does anyone still use it?

The code seems to have been there since the distinction between libc5
and libc6 was introduced in Exim 1.91 in April 1998.

I can't see any objection to dropping support for libc5 now that 7 years
have passed. Anyone object?

> Slightly controversially,

*Very* slightly, IMO!


--
Philip Hazel University of Cambridge Computing Service,
ph10 [at] cus Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book


ph10 at cus

Jun 24, 2005, 3:31 AM

Post #5 of 5 (157 views)
Permalink
Re: Candidate for 4.52 release -- make failure, "No rule to make target `lookups/lookups.a', needed by `exim'" [In reply to]

On Thu, 23 Jun 2005, OpenMacNews wrote:

> tho, next 'make' failure is:
>
> ...
>
> cc exim_monitor/em_version.c
> cc -o eximon.bin
> /usr/bin/ld: warning multiple definitions of symbol __XawTextSearch
> em_TextPop.o definition of __XawTextSearch in section (__TEXT,__text)
> /usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of __XawTextSearch
> /usr/bin/ld: warning multiple definitions of symbol __XawTextDoSearchAction
> em_TextPop.o definition of __XawTextDoSearchAction in section
> (__TEXT,__text)
> /usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of
> __XawTextDoSearchAction
> /usr/bin/ld: warning multiple definitions of symbol
> __XawTextPopdownSearchAction
> em_TextPop.o definition of __XawTextPopdownSearchAction in section
> (__TEXT,__text)
> /usr/X11R6/lib/libXaw.dylib(TextPop.o) definition of
> __XawTextPopdownSearchAction
> >>> exim monitor binary built

Only warnings, so probably OK. The X11 code is ancient. One of these
days eximon should be rewritten to use a modern toolkit. Here is a
comment from OS/Makefile-Default:

# A modified version of the Athena TextPop module is supplied with Exim. The
# modification is to remove the "replace" part of the "search and replace"
# operation because it isn't wanted. TextPop is only one of a number of
# modules that make up the Text widget. Some antique link editors cannot handle
# the case of a replacement module for one of a set of modules. To allow
# the monitor to be linked in such cases, set the value of EXIMON_TEXTPOP
# to be empty. The search operations will then contain a useless "replace"
# option, which is untidy, but does no harm.

EXIMON_TEXTPOP=em_TextPop.o

> cc dnsdb.c
> dnsdb.c:45: error: 'T_A' undeclared here (not in a function)
> dnsdb.c:52: error: 'T_CNAME' undeclared here (not in a function)
> dnsdb.c:54: error: 'T_MX' undeclared here (not in a function)
> dnsdb.c:56: error: 'T_NS' undeclared here (not in a function)
> dnsdb.c:57: error: 'T_PTR' undeclared here (not in a function)

Something needs including to get these macros. On my Linux box these
values are defined in

/usr/include/arpa/nameser_compat.h:

which is included by

/usr/include/arpa/nameser.h

which is included by Exim's "#include <arpa/nameser.h>".

I note that the actual definitions are of the form

#define T_CNAME ns_t_cname

so it looks like a naming change. I will insert some conditionals
definitions to check for the ns_t_ formats and define the old ones
myself if they aren't already defined. Maybe that will solve this one.

> dnsdb.c: In function 'dnsdb_find':
> dnsdb.c:223: warning: comparison between pointer and integer
> dnsdb.c:235: warning: pointer/integer type mismatch in conditional
> expression
> dnsdb.c:235: warning: pointer/integer type mismatch in conditional
> expression
> dnsdb.c:235: warning: initialization makes integer from pointer without a
> cast
> dnsdb.c:243: warning: comparison between pointer and integer
> dnsdb.c:283: warning: comparison between pointer and integer
> dnsdb.c:320: warning: comparison between pointer and integer

These are a consequence of the lack of T_ macros, I think.

--
Philip Hazel University of Cambridge Computing Service,
ph10 [at] cus Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book

exim dev RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.