[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: rpm-4.1-0.56 packages available

Hi

I'm still trying to get this to compile on darwin with no luck. I seen this announcement, so I thought i'd try the latest version. There is still a problem with the compile script in the db3 directory, with the tab not being put into the Makefile. I just added a vi Makefile command before the make command, and I can edit the Makefile to correct everything, so it all configures well. But when I get to compiling the beecrypt stuff, I get problems:

/bin/sh ./libtool --mode=compile cc -DHAVE_CONFIG_H -I. -I. -I. -DOPTIMIZE_POWERPC -O3 -g -I/System/Library/Frameworks/JavaVM.framework/Versions/1.3.1/Home/../../../Headers -c -o entropy.lo `test -f 'entropy.c' || echo './'`entropy.c
rm -f .libs/entropy.lo
cc -DHAVE_CONFIG_H -I. -I. -I. -DOPTIMIZE_POWERPC -O3 -g -I/System/Library/Frameworks/JavaVM.framework/Versions/1.3.1/Home/../../../Headers -c entropy.c -fno-common -DPIC -o .libs/entropy.lo
entropy.c:1104: undefined or invalid # directive
entropy.c: In function `entropy_randombits':
entropy.c:986: storage size of `my_aiocb' isn't known
entropy.c:994: sizeof applied to an incomplete type
entropy.c:998: `SIGEV_NONE' undeclared (first use in this function)
entropy.c:998: (Each undeclared identifier is reported only once
entropy.c:998: for each function it appears in.)
entropy.c:1044: `AIO_ALLDONE' undeclared (first use in this function)
entropy.c:1048: `AIO_NOTCANCELED' undeclared (first use in this function)
make[2]: *** [entropy.lo] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2


I don't remember this happening in the compile of the last beta of this, but I kind of trashed that directory before i pulled the new one off the ftp server. It would also be nice if you would put a tar.gz of the source up.

I also tried to compile under freebsd 4.6. Pretty similar problems. It doesn't like the missing tab in the db3/Makefile It likes to stop even earlier when compiling the beecrypt stuff:

Making all in tests
/bin/bash ./libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I. -I. -pthread -DOPTIMIZE_I386 -pthread -O3 -g -march=i386 -mcpu=i686 -fomit-frame-pointer -I -c -o aes.lo `test -f 'aes.c' || echo './'`aes.c
mkdir .libs
gcc -DHAVE_CONFIG_H -I. -I. -I. -pthread -DOPTIMIZE_I386 -pthread -O3 -g -march=i386 -mcpu=i686 -fomit-frame-pointer -I -c aes.c -fPIC -DPIC -o .libs/aes.lo
In file included from system.h:73,
from aes.c:27:
/usr/include/malloc.h:2: warning: #warning "this file includes <malloc.h> which is deprecated, use <stdlib.h> instead"
/tmp/ccDL9ObO.o: In function `aesEncrypt':
/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.c(.text+0x4a5): undefined reference to `swapu32'
/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.c(.text+0x4bd): undefined reference to `swapu32'
/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.c(.text+0x4d5): undefined reference to `swapu32'
/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.c(.text+0x4ea): undefined reference to `swapu32'
/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.c(.text+0x18c1): undefined reference to `swapu32'
/tmp/ccDL9ObO.o(.text+0x18d0):/home/users/j/ji/jimmacr/rpm-4.1/beecrypt/aes.
c: more undefined references to `swapu32' follow
/usr/lib/crt1.o: In function `_start':
/usr/lib/crt1.o(.text+0x79): undefined reference to `main'
*** Error code 1

Stop in /home/users/j/ji/jimmacr/rpm-4.1/beecrypt.
*** Error code 1


On Friday, July 26, 2002, at 11:43 AM, Jeff Johnson wrote:


There rpm-4.1-0.56 packages available for testing at

ftp://ftp.rpm.org/pub/rpm/test-4.1

I've included source and binary packages compiled for all Red Hat platforms.
All the packages are signed with the BETA-GPG-KEY, also present at the
above URL.

The point of this test release is to make a reference starting point available
for testing on previous Red Hat platforms, and to get feedback on non-linux
portability issues. Bug reports, comments, RFE's etc at
http://bugzilla.redhat.com
please. No testing at all has been done on the 7x/6x packages so far.

Since the previous rpm-4.1 test release, the following features have been
added:

1) Header signatures and digests can (if configured) be verified when
   retrieved from the rpm database. See /usr/lib/rpm/macros for comments
   about how to configure %_vsflags_install et al macros.

Note: this is only mechanism, a better/configurable security policy
is gonna be needed in rpm somewhen soon. Suggestions/patches/audits are
very much appreciated.

2) The rpm database permits concurrent access. That means that it is now
   possible to run rpm in %post scriptlets.

   Note: What still remains is to find out whether there are deadlocks
   (there are), and whether the deadlocks can be avoided or otherwise
   handled gracefully. I'd really like to support (at least read)
   concurrent access to the rpm database, but it's gonna take a lot
   of careful (i.e. reproducible) testing to achieve that goal. Any
   and all help is appreciated. What's very promising is that the
   problems are deadlocks, not segfaults, but reproducing deadlocks
   is gonna be quite challenging.

3) The rpmdb-redhat package (which contains an "everything" rpm database),
if installed, will be used to provide suggested solutions for unresolved
dependencies. Try installing the rpmdb-redhat package from Raw Hide if
interested.

4) /usr/lib/rpm/rpmcache has been added to the rpm-devel package. rpmcache
can be used to create a custom rpm database to provide suggested solutions
for unresolved dependencies. See rpmcache(8) for some hints about how to
use.

5) /usr/bin/rpmgraph has been added to the rpm-devel package. rpmgrapgh
will generate dependsnecy graphs of packages for display by graphwiz.
See rpmgraph(8) for details. If nowhere else, there's a graphvix source
rpm that can be compiled to display the output at
ftp://people.redhat.com/jbj/graphviz-1.7.14-1.src.rpm

There are several goals in rpm-4.1:

1) new, header-only, digital signatures and digests for packages.

2) verifying header signatures/digests when the header is read.

3) replacing the internal representation of headers with smaller data
   structures to reduce memory footprint.

4) enriching and stabilizing the rpmlib API.

Both 3) and 4) are gonna be quite painful for developers, as the
entire rpmlib API has changed. What's good is that I can now expose
more of the internals of rpmlib more or less opaquely.

What's bad is that everything has changed. I'll be happy to answer questions
or help with porting (assuming time available) as needed.





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []