[Monotone-devel] PATCH: note_netsync_revision_received apparently missing author cert

[Bruce Stephens]

Someone on #monotone noted that their hook didn't seem to notice the
author cert.  They had a loop using ipairs rather than pairs, which
seems reasonable...

Except that tables in Lua count from 1, not 0.  I don't know
why---seems an odd choice to me.  (And IIUC, it's by convention of the
various functions like ipairs, etc., not something more fundamental.)

I haven't checked the other hooks, but this isn't the first time it's
happened, so I suggest using pairs rather than ipairs, or at least
trying that.  I suspect other hooks receiving tables will receive
0-indexed things, and if you use ipairs, you'll miss one.

So my (untested) guess is there's a bug in lua_hooks.cc:

#
# old_revision [fce69d16cb214674c459f66465ce6b62a272ff23]
#
# patch "lua_hooks.cc"
#  from [9f844c1bf6692f8bb488e5a596caa2f622f8a454]
#    to [6359909e6058a18614b49e0a02967d5725fdb10e]
#
============================================================
--- lua_hooks.cc        9f844c1bf6692f8bb488e5a596caa2f622f8a454
+++ lua_hooks.cc        6359909e6058a18614b49e0a02967d5725fdb10e
@@ -841,7 +841,7 @@ lua_hooks::hook_note_netsync_revision_re
   int n=0;
   for (cdat::const_iterator i = certs.begin(); i != certs.end(); ++i)
     {
-      ll.push_int(n++);
+      ll.push_int(++n);
       ll.push_table();
       ll.push_str("key");
       ll.push_str(i->first());