>>> should only be built with execv() if the cvs plugin is loaded?
>>> --[[schmonz]]
+>>>> Hadn't considered that. While in wrapper mode the normal getopt is not
+>>>> done, plugin getopt still runs, and so any unsafe options that
+>>>> other plugins support could be a problem if another user runs
+>>>> the setuid wrapper and passes those options through. --[[Joey]]
+
+>>>>> I've tried compiling the argument check into the wrapper as
+>>>>> the first thing main() does, and was surprised to find that
+>>>>> this doesn't prevent the `cvs add <dir>` deadlock in a web
+>>>>> commit. I was convinced this'd be a reasonable solution,
+>>>>> especially if conditionalized on the cvs plugin being loaded,
+>>>>> but it doesn't work. And I stuck debug printfs at the beginning
+>>>>> of all the rcs_foo() subs, and whatever `cvs add <dir>` is
+>>>>> doing to ikiwiki isn't visible to my plugin, because none of
+>>>>> those subs are getting called. Nuts. Can you think of anything
+>>>>> else that might solve the problem, or should I go back to
+>>>>> generating a minimal wrapper wrapper that checks for just
+>>>>> this one thing? --[[schmonz]]
+
+>>>>>> I don't see how there could possibly be a difference between
+>>>>>> ikiwiki's C wrapper and your shell wrapper wrapper here. --[[Joey]]
+
+>>>>>>> I was comparing strings overly precisely. Fixed on my branch.
+>>>>>>> I've also knocked off the two most pressing to-do items. I
+>>>>>>> think the plugin's ready for prime time. --[[schmonz]]
+
> Thing 2 I'm less sure of. (I'd like to see the web UI return
> immediately on save anyway, to a temporary "rebuilding, please wait
> if you feel like knowing when it's done" page, but this problem
>>> if a configured post-commit hook is missing, and it seems fine,
>>> probably also thanks to IPC::Cmd.
>>> --[[schmonz]]
+
+----
+
+
+Further review.. --[[Joey]]
+
+I don't understand what `cvs_shquote_commit` is
+trying to do with the test message, but it seems
+highly likely to be insecure; I never trust anything
+that relies on safely quoting user input passed to the shell.
+
+(As an aside, `shell_quote` can die on certian inputs.)
+
+Seems to me that, if `IPC::Cmd` exposes input to the shell
+(which I have not verified but its docs don't specify; a bad sign)
+you chose the wrong tool and ended up doing down the wrong
+route, dragging in shell quoting problems and fixes. Since you
+chose to use `IPC::Cmd` just because you wanted to shut
+up CVS stderr, my suggestion would be to use plain `system`
+to run the command, with stderr temporarily sent to /dev/null:
+
+ open(my $savederr, ">&STDERR");
+ open(STDERR, ">", "/dev/null");
+ my $ret=system("cvs", "-Q", @_);
+ open(STDERR, ">$savederr");
+
+`cvs_runcvs` should not take an array reference. It's
+usual for this type of function to take a list of parameters
+to pass to the command.
+
+> Thanks for reading carefully. I've tested your suggestions and
+> applied them on my branch. --[[schmonz]]
+
+----
+
+I've abstracted out CVS's involvement in the wrapper, adding a new
+"wrapperargcheck" hook to examine `argc/argv` and return success or
+failure (failure causes the wrapper to terminate) and implementing
+this hook in the plugin. In the non-CVS case, the check immediately
+returns success, so the added overhead is just a function call.
+
+Given how rarely anything should need to reach in and modify the
+wrapper -- I'd go so far as to say we shouldn't make it too easy
+-- I don't think it's worth the effort to try and design a more
+general-purpose way to do so. If and when some other problem thinks
+it wants to be solved by a new wrapper hook, it's easy enough to add
+one. Until then, I'd say it's more important to keep the wrapper as
+short and clear as possible. --[[schmonz]]