OK, I packaged up the current CVS as our first preview release for 2.6.0. You can grab it here: http://samba.org/ftp/rsync/preview/rsync-2.6.0pre1.tar.gz The MD5 checksum is: 70e9dea967f083c231b7821ef35aef1b rsync-2.6.0pre1.tar.gz There is not currently a .sig file for the package, but I'm looking into that next. Please test this and let me know if we have any remaining issues to work through before the final release. Thanks, ..wayne..
On Fri 19 Dec 2003, Wayne Davison wrote:> Please test this and let me know if we have any remaining issues to work > through before the final release. Thanks,One tiny little nit, when compiling on alpha, I get the following warning: alpha-linux-gcc -I. -I. -Wall -O2 -c sender.c -o sender.o sender.c: In function `read_sum_head': sender.c:51: warning: int format, different type arg (arg 3) It's a size_t being printed as an int. Here's a patch: --- sender.c.orig 2003-12-19 23:08:13.000000000 +0100 +++ sender.c 2003-12-19 23:08:21.000000000 +0100 @@ -47,8 +47,8 @@ } else { sum->s2length = read_int(f); if (sum->s2length > MD4_SUM_LENGTH) { - rprintf(FERROR, "Invalid checksum length %d\n", - sum->s2length); + rprintf(FERROR, "Invalid checksum length %ld\n", + (long)sum->s2length); exit_cleanup(RERR_PROTOCOL); } } Alternatively, it could have been cast to a normal int and the format left as %d. However, I thought that this was the safest, as although I can't really see how sum->s2length could contain more than 32 bits, I guess it was declared as size_t for some reason... Paul Slootman
On Fri 19 Dec 2003, Wayne Davison wrote:> OK, I packaged up the current CVS as our first preview release for > 2.6.0. You can grab it here:I've built i386 and alpha Debian packages, uploaded to experimental. They're also available at http://wurtel.op.het.net/rsync_2.5.99pre1-1_alpha.deb and http://wurtel.op.het.net/rsync_2.5.99pre1-1_i386.deb Sparc maybe to follow tomorrow, if there's any demand... (my sparc's very slow and noisy...) The different numbering is because in Debian, 2.6.0pre1 is considered higher than 2.6.0 by the package management tools, and I don't want to screw up the version of the real release... The dependencies are: libc6 (>= 2.3.2.ds1-4), libpopt0 (>= 1.7) An up to date "testing" should suffice. The source package is also available in the Debian experimental archive, and alternatively at http://wurtel.op.het.net/rsync_2.5.99pre1-1.dsc http://wurtel.op.het.net/rsync_2.5.99pre1-1.diff.gz http://wurtel.op.het.net/rsync_2.5.99pre1.orig.tar.gz The last is identical to:> http://samba.org/ftp/rsync/preview/rsync-2.6.0pre1.tar.gzIt's just been renamed to comply with the versioning thing. Paul Slootman -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: Digital signature Url : http://lists.samba.org/archive/rsync/attachments/20031219/63db2dc4/attachment.bin
On Fri, 19 Dec 2003, Wayne Davison <wayned@samba.org> wrote:> > OK, I packaged up the current CVS as our first preview release for > 2.6.0. You can grab it here: > > http://samba.org/ftp/rsync/preview/rsync-2.6.0pre1.tar.gzI checked the sources to see if the various fixes that I've been using are in 2.6.0. They are, but NEWS doesn't mention two of them. Add: * Fixed bogus "malformed address {hostname}" message in rsyncd log when checking IP address against hostnames from "hosts allow" and "hosts deny" parameters in config file. * Print heap statistics when verbose >= 2 instead of when >= 1 (I suppose the 1st item could be shortened to just the first line.) In the original patch to main.c that I submitted for the 2nd item, this got left out when it was applied: < rprintf(FINFO, RSYNC_NAME "[%d] (%s%s%s) heap statistics:\n", ---> rprintf(FINFO, "\n" RSYNC_NAME "[%d] (%s%s%s) heap statistics:\n",That leading "\n" separates the heap statistics from the filelist and other verbose output. Can you please add that? Also, --version says: Copyright (C) 1996-2002 by Andrew Tridgell and others Change 2002 to 2003. If another 12 days go by, change it to 2004. :) A suggestion - at the end of the ./configure output, print: ******************************************************************** * As of v2.6.0 the default remote shell is ssh instead of rsh. * * To use previous default of rsh: ./configure --with-rsh=rsh * ******************************************************************** to make it really clear about the new default. -- John Van Essen Univ of MN Alumnus <vanes002@umn.edu>