Martin Pool [Mon, 8 Apr 2002 04:16:46 +0000 (04:16 +0000)]
Split code out into separate files and remove some global variables to
reduce symbol dependencies between files and therefore make it easier
to write unit tests. The diff is large, but the actual code changes
are pretty small.
Martin Pool [Mon, 8 Apr 2002 04:12:31 +0000 (04:12 +0000)]
Little stub module that replaces functions such as rprintf called from
many places with a redirector to stderr. If you just want to test a
particular routine you can link to this rather than the whole framework.
Martin Pool [Mon, 8 Apr 2002 04:10:20 +0000 (04:10 +0000)]
Split code out into separate files and remove some global variables to
reduce symbol dependencies between files and therefore make it easier
to write unit tests. The diff is large, but the actual code changes
are pretty small.
Martin Pool [Wed, 3 Apr 2002 05:55:54 +0000 (05:55 +0000)]
Change code that writes s_count to a batch file to squish it into an
int first -- a stopgap for compilers that reasonably complain about a
ternary that returns either a size_t or an int.
Really we should not write an int, but rather a larger type. Somebody
more familiar with the batch code should look at it.
Martin Pool [Wed, 3 Apr 2002 02:33:42 +0000 (02:33 +0000)]
Commit patch based on one from John L. Allen to fix "forward name
lookup failed" on AIX 4.3.3.
When doing an name->addr lookup on the client address, there's no
point including the service name, because it can't be spoofed and it
apparently causes trouble when not present in /etc/services.
Martin Pool [Wed, 27 Mar 2002 05:10:44 +0000 (05:10 +0000)]
If configured with --enable-maintainer-mode, then on receipt of a
fatal signal rsync will try to open an xterm running gdb, similarly to
Samba's "panic action" or GNOME's bug-buddy.
Martin Pool [Tue, 26 Mar 2002 10:28:57 +0000 (10:28 +0000)]
Fix suggested by Jos for hands.test. Some of the functionality
previously built into Phil's test was clashing with the runtests.sh
framework. I think one of the core problems might be that scripts
which don't explicitly "exit 0" have return codes which vary depending
on the particular shell. (c.f. the autoconf portability guide.)
This might break the Debian test suite but I'm afraid I don't know
enough about how it works yet. Hopefully eventually it can become
just a wrapper around our 'make check'.
Martin Pool [Fri, 22 Mar 2002 05:06:36 +0000 (05:06 +0000)]
Use `id -G` rather than `groups` to get a list of groups for this
user, so that we can cope on systems like Cygwin that tend to have
group names containing spaces.
Martin Pool [Wed, 20 Mar 2002 01:09:49 +0000 (01:09 +0000)]
Patch from Paul Haas:
* Fix situation where failure to fork (e.g. because out of process
slots) would cause rsync to kill all processes owned by the
current user. Yes, really! (Paul Haas, Martin Pool)
Also, add a little more documentation and paranoia to make sure we
never call kill(2) with a negative pid.
David Dykstra [Tue, 19 Mar 2002 20:16:42 +0000 (20:16 +0000)]
Now that whole_file and no_whole_file are separate variables, have the
--whole-file and --no-whole-file options set/reset both variables so if
more than one option is specified the last one will be the one honored.