Don't check _accept_mutation in _get_mut_with_use.

Most of the code that was replaced by _get_mut_with_use did not care
whether the mutation was acceptible, and some of the code was broken by
the change.  In particular, _delete_random_slime_mutation indexed
you.mutations directly with the result, causing a crash when the
selected mutation was unacceptible [1]---and even if the crash were
fixed, the behaviour would be wrong because it would never select a
maxed-out slime mutation.

Move the _accept_mutation call out of _get_mut_with_use and into the one
caller that needed it, _get_random_mutation.

[1] For example,
http://crawl.akrasiac.org/rawdata/DreamWalker/crash-DreamWalker-20140611-033210.txt

1 files changed, 2 insertions, 4 deletions

diff --git a/crawl-ref/source/mutation.cc b/crawl-ref/source/mutation.cc
index c3e47da988..4ed32550d5 100644
--- a/crawl-ref/source/mutation.cc
+++ b/crawl-ref/source/mutation.cc
@@ -995,9 +995,7 @@ static mutation_type _get_mut_with_use(mut_use_type mt)
         if (cweight >= 0)
             continue;
 
-        if (_accept_mutation(mut_data[i].mutation, true))
-            return mut_data[i].mutation;
-        return NUM_MUTATIONS;
+        return mut_data[i].mutation;
     }
 
     die("Error while selecting mutations");
@@ -1086,7 +1084,7 @@ static mutation_type _get_random_mutation(mutation_type mutclass)
     for (int attempt = 0; attempt < 100; ++attempt)
     {
         mutation_type mut = _get_mut_with_use(mt);
-        if (mut != NUM_MUTATIONS)
+        if (_accept_mutation(mut, true))
             return mut;
     }