From 2de0e86e96b24ef5b5fa6c8b846f58c26caa820e Mon Sep 17 00:00:00 2001
From: peterychang <petchang.microsoft.com>
Date: Thu, 2 Jan 2020 14:56:06 -0500
Subject: [PATCH 1/3] deprecate last and pop, implement insert, erase, capacity

---
 python/pylibvw.cc                             | 12 ++--
 vowpalwabbit/autolink.cc                      |  2 +-
 vowpalwabbit/comp_io.cc                       |  2 +-
 vowpalwabbit/conditional_contextual_bandit.cc |  8 +--
 vowpalwabbit/csoaa.cc                         |  4 +-
 vowpalwabbit/ect.cc                           |  2 +-
 vowpalwabbit/example.h                        |  2 +-
 vowpalwabbit/ezexample.h                      |  2 +-
 vowpalwabbit/interact.cc                      | 18 +++---
 vowpalwabbit/io_buf.cc                        |  2 +-
 vowpalwabbit/io_buf.h                         |  5 +-
 vowpalwabbit/kernel_svm.cc                    |  8 +--
 vowpalwabbit/label_dictionary.cc              |  4 +-
 vowpalwabbit/lda_core.cc                      |  4 +-
 vowpalwabbit/memory_tree.cc                   |  4 +-
 vowpalwabbit/mwt.cc                           |  3 +-
 vowpalwabbit/nn.cc                            |  2 +-
 vowpalwabbit/parse_example.cc                 |  9 +--
 vowpalwabbit/parser.cc                        | 13 ++--
 vowpalwabbit/recall_tree.cc                   |  2 +-
 vowpalwabbit/search.cc                        | 14 ++---
 vowpalwabbit/search_dep_parser.cc             | 40 ++++++-------
 vowpalwabbit/search_entityrelationtask.cc     |  4 +-
 vowpalwabbit/search_graph.cc                  |  2 +-
 vowpalwabbit/slim/src/vw_slim_predict.cc      |  4 +-
 vowpalwabbit/v_array.h                        | 59 +++++++++++++++----
 vowpalwabbit/vwdll.cpp                        |  2 +-
 27 files changed, 138 insertions(+), 95 deletions(-)

diff --git a/python/pylibvw.cc b/python/pylibvw.cc
index 6c4f09dcbd3..97fed70ca5f 100644
--- a/python/pylibvw.cc
+++ b/python/pylibvw.cc
@@ -381,11 +381,12 @@ void ex_push_dictionary(example_ptr ec, vw_ptr vw, py::dict& dict)
 
 bool ex_pop_feature(example_ptr ec, unsigned char ns)
 { if (ec->feature_space[ns].size() == 0) return false;
-  float val = ec->feature_space[ns].values.pop();
+  float val = ec->feature_space[ns].values.back();
+  ec->feature_space[ns].values.pop_back();
   if (ec->feature_space[ns].indicies.size()> 0)
-    ec->feature_space[ns].indicies.pop();
+    ec->feature_space[ns].indicies.pop_back();
   if (ec->feature_space[ns].space_names.size()> 0)
-    ec->feature_space[ns].space_names.pop();
+    ec->feature_space[ns].space_names.pop_back();
   ec->num_features--;
   ec->feature_space[ns].sum_feat_sq -= val * val;
   ec->total_sum_feat_sq -= val * val;
@@ -401,7 +402,8 @@ void ex_erase_namespace(example_ptr ec, unsigned char ns)
 
 bool ex_pop_namespace(example_ptr ec)
 { if (ec->indices.size() == 0) return false;
-  unsigned char ns = ec->indices.pop();
+  unsigned char ns = ec->indices.back();
+  ec->indices.pop_back();
   ex_erase_namespace(ec, ns);
   return true;
 }
@@ -446,7 +448,7 @@ void unsetup_example(vw_ptr vwP, example_ptr ae)
     if (hit_constant >= 0)
     { for (size_t i=hit_constant; i<N-1; i++)
         ae->indices[i] = ae->indices[i+1];
-      ae->indices.pop();
+      ae->indices.pop_back();
     }
   }
 
diff --git a/vowpalwabbit/autolink.cc b/vowpalwabbit/autolink.cc
index 988c34933dd..7c7734334f6 100644
--- a/vowpalwabbit/autolink.cc
+++ b/vowpalwabbit/autolink.cc
@@ -74,7 +74,7 @@ void VW::autolink::reset_example(example& ec)
   features& fs = ec.feature_space[autolink_namespace];
   ec.total_sum_feat_sq -= fs.sum_feat_sq;
   fs.clear();
-  ec.indices.pop();
+  ec.indices.pop_back();
 }
 
 template <bool is_learn>
diff --git a/vowpalwabbit/comp_io.cc b/vowpalwabbit/comp_io.cc
index cd784d0180f..f19e5dae6a1 100644
--- a/vowpalwabbit/comp_io.cc
+++ b/vowpalwabbit/comp_io.cc
@@ -84,7 +84,7 @@ bool comp_io_buf::close_file()
     gzclose(gz_files.back());
     gz_files.pop_back();
     if (!files.empty())
-      files.pop();
+      files.pop_back();
     return true;
   }
   return false;
diff --git a/vowpalwabbit/conditional_contextual_bandit.cc b/vowpalwabbit/conditional_contextual_bandit.cc
index a040a38a739..93fbb0aef0d 100644
--- a/vowpalwabbit/conditional_contextual_bandit.cc
+++ b/vowpalwabbit/conditional_contextual_bandit.cc
@@ -250,13 +250,13 @@ void inject_slot_id(ccb& data, example* shared, size_t id)
 template <bool audit>
 void remove_slot_id(example* shared)
 {
-  shared->feature_space[ccb_id_namespace].indicies.pop();
-  shared->feature_space[ccb_id_namespace].values.pop();
-  shared->indices.pop();
+  shared->feature_space[ccb_id_namespace].indicies.pop_back();
+  shared->feature_space[ccb_id_namespace].values.pop_back();
+  shared->indices.pop_back();
 
   if (audit)
   {
-    shared->feature_space[ccb_id_namespace].space_names.pop();
+    shared->feature_space[ccb_id_namespace].space_names.pop_back();
   }
 }
 
diff --git a/vowpalwabbit/csoaa.cc b/vowpalwabbit/csoaa.cc
index 22ffafeec8f..892089b401c 100644
--- a/vowpalwabbit/csoaa.cc
+++ b/vowpalwabbit/csoaa.cc
@@ -228,7 +228,7 @@ void unsubtract_example(example* ec)
     return;
   }
 
-  if (ec->indices.last() != wap_ldf_namespace)
+  if (ec->indices.back() != wap_ldf_namespace)
   {
     std::cerr
         << "internal error (bug): trying to unsubtract_example, but either it wasn't added, or something was added "
@@ -241,7 +241,7 @@ void unsubtract_example(example* ec)
   ec->num_features -= fs.size();
   ec->total_sum_feat_sq -= fs.sum_feat_sq;
   fs.clear();
-  ec->indices.decr();
+  ec->indices.pop_back();
 }
 
 void make_single_prediction(ldf& data, single_learner& base, example& ec)
diff --git a/vowpalwabbit/ect.cc b/vowpalwabbit/ect.cc
index cbfc786b36a..af2a061c2ea 100644
--- a/vowpalwabbit/ect.cc
+++ b/vowpalwabbit/ect.cc
@@ -174,7 +174,7 @@ size_t create_circuit(ect& e, uint64_t max_label, uint64_t eliminations)
           e.directions[direction_index].loser = 0;
       }
       if (tournaments[t].size() % 2 == 1)
-        new_tournaments[t].push_back(tournaments[t].last());
+        new_tournaments[t].push_back(tournaments[t].back());
     }
     e.all_levels.push_back(new_tournaments);
     level++;
diff --git a/vowpalwabbit/example.h b/vowpalwabbit/example.h
index b1ea65c8872..02e787e70e4 100644
--- a/vowpalwabbit/example.h
+++ b/vowpalwabbit/example.h
@@ -103,7 +103,7 @@ inline int example_is_newline(example const& ec)
 {  // if only index is constant namespace or no index
   if (!ec.tag.empty())
     return false;
-  return ((ec.indices.empty()) || ((ec.indices.size() == 1) && (ec.indices.last() == constant_namespace)));
+  return ((ec.indices.empty()) || ((ec.indices.size() == 1) && (ec.indices.back() == constant_namespace)));
 }
 
 inline bool valid_ns(char c) { return !(c == '|' || c == ':'); }
diff --git a/vowpalwabbit/ezexample.h b/vowpalwabbit/ezexample.h
index 5e958a1cee7..e043385f50d 100644
--- a/vowpalwabbit/ezexample.h
+++ b/vowpalwabbit/ezexample.h
@@ -177,7 +177,7 @@ class ezexample
 
       current_seed = past_seeds.back();
       past_seeds.pop_back();
-      ec->indices.pop();
+      ec->indices.pop_back();
       example_changed_since_prediction = true;
     }
   }
diff --git a/vowpalwabbit/interact.cc b/vowpalwabbit/interact.cc
index 3d9786cf6e9..7812dbf612a 100644
--- a/vowpalwabbit/interact.cc
+++ b/vowpalwabbit/interact.cc
@@ -123,14 +123,13 @@ void predict_or_learn(interact& in, LEARNER::single_learner& base, example& ec)
     std::cout<< std::endl;*/
 
   // remove 2nd namespace
-  int n2_i = -1;
-  for (size_t i = 0; i < ec.indices.size(); i++)
+  size_t n2_i = 0;
+  size_t indices_original_size = ec.indices.size();
+  for (; n2_i < indices_original_size; ++n2_i)
   {
-    if (ec.indices[i] == in.n2)
+    if (ec.indices[n2_i] == in.n2)
     {
-      n2_i = (int)i;
-      memmove(&ec.indices[n2_i], &ec.indices[n2_i + 1], sizeof(unsigned char) * (ec.indices.size() - n2_i - 1));
-      ec.indices.decr();
+      ec.indices.erase(n2_i);
       break;
     }
   }
@@ -139,10 +138,9 @@ void predict_or_learn(interact& in, LEARNER::single_learner& base, example& ec)
   if (is_learn)
     base.learn(ec);
 
-  // re-insert namespace into the right position
-  ec.indices.incr();
-  memmove(&ec.indices[n2_i + 1], &ec.indices[n2_i], sizeof(unsigned char) * (ec.indices.size() - n2_i - 1));
-  ec.indices[n2_i] = in.n2;
+  // re-insert namespace into the right position (if one was removed)
+  if (n2_i < indices_original_size)
+    ec.indices.insert(n2_i, in.n2);
 
   f1.deep_copy_from(in.feat_store);
   ec.total_sum_feat_sq = in.total_sum_feat_sq;
diff --git a/vowpalwabbit/io_buf.cc b/vowpalwabbit/io_buf.cc
index 4baf0829176..996e3c108c5 100644
--- a/vowpalwabbit/io_buf.cc
+++ b/vowpalwabbit/io_buf.cc
@@ -103,7 +103,7 @@ void io_buf::buf_write(char*& pointer, size_t n)
       flush();
     else  // Array is short, so increase size.
     {
-      space.resize(2 * (space.end_array - space.begin()));
+      space.resize(2 * space.capacity());
       space.end() = space.begin();
       head = space.begin();
     }
diff --git a/vowpalwabbit/io_buf.h b/vowpalwabbit/io_buf.h
index 6cadafbf34d..5f1718a9bad 100644
--- a/vowpalwabbit/io_buf.h
+++ b/vowpalwabbit/io_buf.h
@@ -172,7 +172,7 @@ class io_buf
     if (space.end_array - space.end() == 0)
     {  // reallocate to twice as much space
       size_t head_loc = head - space.begin();
-      space.resize(2 * (space.end_array - space.begin()));
+      space.resize(2 * space.capacity());
       head = space.begin() + head_loc;
     }
     // read more bytes from file up to the remaining allocated space
@@ -204,7 +204,8 @@ class io_buf
   {
     if (!files.empty())
     {
-      close_file_or_socket(files.pop());
+      close_file_or_socket(files.back());
+      files.pop_back();
       return true;
     }
     return false;
diff --git a/vowpalwabbit/kernel_svm.cc b/vowpalwabbit/kernel_svm.cc
index fe396b2094c..f48fcebb491 100644
--- a/vowpalwabbit/kernel_svm.cc
+++ b/vowpalwabbit/kernel_svm.cc
@@ -523,9 +523,9 @@ int remove(svm_params& params, size_t svi)
   }
   svi_e->~svm_example();
   free(svi_e);
-  model->support_vec.pop();
-  model->alpha.pop();
-  model->delta.pop();
+  model->support_vec.pop_back();
+  model->alpha.pop_back();
+  model->delta.pop_back();
   model->num_support--;
   // shift cache
   int alloc = 0;
@@ -536,7 +536,7 @@ int remove(svm_params& params, size_t svi)
     if (svi < rowsize)
     {
       for (size_t i = svi; i < rowsize - 1; i++) e->krow[i] = e->krow[i + 1];
-      e->krow.pop();
+      e->krow.pop_back();
       alloc -= 1;
     }
   }
diff --git a/vowpalwabbit/label_dictionary.cc b/vowpalwabbit/label_dictionary.cc
index 3eeeb79f01c..18c537db5e1 100644
--- a/vowpalwabbit/label_dictionary.cc
+++ b/vowpalwabbit/label_dictionary.cc
@@ -17,8 +17,8 @@ void del_example_namespace(example& ec, namespace_index ns, features& fs)
   features& del_target = ec.feature_space[(size_t)ns];
   assert(del_target.size() >= fs.size());
   assert(ec.indices.size() > 0);
-  if (ec.indices.last() == ns && ec.feature_space[(size_t)ns].size() == fs.size())
-    ec.indices.pop();
+  if (ec.indices.back() == ns && ec.feature_space[(size_t)ns].size() == fs.size())
+    ec.indices.pop_back();
   ec.total_sum_feat_sq -= fs.sum_feat_sq;
   ec.num_features -= fs.size();
   del_target.truncate_to(del_target.size() - fs.size());
diff --git a/vowpalwabbit/lda_core.cc b/vowpalwabbit/lda_core.cc
index e8ce93a625e..379a3ee6b60 100644
--- a/vowpalwabbit/lda_core.cc
+++ b/vowpalwabbit/lda_core.cc
@@ -909,7 +909,7 @@ void learn_batch(lda &l)
   eta = l.all->eta * l.powf((float)l.example_t, -l.all->power_t);
   minuseta = 1.0f - eta;
   eta *= l.lda_D / batch_size;
-  l.decay_levels.push_back(l.decay_levels.last() + log(minuseta));
+  l.decay_levels.push_back(l.decay_levels.back() + log(minuseta));
 
   l.digammas.clear();
   float additional = (float)(l.all->length()) * l.lda_rho;
@@ -1259,7 +1259,7 @@ void end_examples(lda &l, T &weights)
   for (typename T::iterator iter = weights.begin(); iter != weights.end(); ++iter)
   {
     float decay_component =
-        l.decay_levels.last() - l.decay_levels.end()[(int)(-1 - l.example_t + (&(*iter))[l.all->lda])];
+        l.decay_levels.back() - l.decay_levels.end()[(int)(-1 - l.example_t + (&(*iter))[l.all->lda])];
     float decay = fmin(1.f, correctedExp(decay_component));
 
     weight *wp = &(*iter);
diff --git a/vowpalwabbit/memory_tree.cc b/vowpalwabbit/memory_tree.cc
index e186440431c..cbac129075f 100644
--- a/vowpalwabbit/memory_tree.cc
+++ b/vowpalwabbit/memory_tree.cc
@@ -33,14 +33,14 @@ void remove_at_index(v_array<T>& array, uint32_t index)
   }
   if (index == array.size() - 1)
   {
-    array.pop();
+    array.pop_back();
     return;
   }
   for (size_t i = index + 1; i < array.size(); i++)
   {
     array[i - 1] = array[i];
   }
-  array.pop();
+  array.pop_back();
   return;
 }
 
diff --git a/vowpalwabbit/mwt.cc b/vowpalwabbit/mwt.cc
index 3f6e2acd251..3ed93d5cd3b 100644
--- a/vowpalwabbit/mwt.cc
+++ b/vowpalwabbit/mwt.cc
@@ -130,7 +130,8 @@ void predict_or_learn(mwt& c, single_learner& base, example& ec)
   if (exclude || learn)
     while (!c.indices.empty())
     {
-      unsigned char ns = c.indices.pop();
+      unsigned char ns = c.indices.back();
+      c.indices.pop_back();
       std::swap(c.feature_space[ns], ec.feature_space[ns]);
     }
 
diff --git a/vowpalwabbit/nn.cc b/vowpalwabbit/nn.cc
index f32bb02d0ac..8be94b55e8b 100644
--- a/vowpalwabbit/nn.cc
+++ b/vowpalwabbit/nn.cc
@@ -320,7 +320,7 @@ void predict_or_learn_multi(nn& n, single_learner& base, example& ec)
       ec.total_sum_feat_sq -= tmp_sum_feat_sq;
       ec.feature_space[nn_output_namespace].sum_feat_sq = 0;
       std::swap(ec.feature_space[nn_output_namespace], save_nn_output_namespace);
-      ec.indices.pop();
+      ec.indices.pop_back();
     }
     else
     {
diff --git a/vowpalwabbit/parse_example.cc b/vowpalwabbit/parse_example.cc
index 4167795cb9c..965ec0d27f3 100644
--- a/vowpalwabbit/parse_example.cc
+++ b/vowpalwabbit/parse_example.cc
@@ -219,7 +219,7 @@ class TC_parser
             d = '.';
           else
             d = '#';
-          // if ((spelling.size() == 0) || (spelling.last() != d))
+          // if ((spelling.size() == 0) || (spelling.back() != d))
           _spelling.push_back(d);
         }
 
@@ -444,10 +444,11 @@ void substring_to_example(vw* all, example* ae, VW::string_view example)
     std::vector<VW::string_view> tokenized;
     tokenize(' ', label_space, all->p->words);
     if (all->p->words.size() > 0 &&
-        (all->p->words.last().end() == label_space.end() ||
-        all->p->words.last().front() == '\''))  // The last field is a tag, so record and strip it off
+        (all->p->words.back().end() == label_space.end() ||
+            all->p->words.back().front() == '\''))  // The last field is a tag, so record and strip it off
     {
-      VW::string_view tag = all->p->words.pop();
+      VW::string_view tag = all->p->words.back();
+      all->p->words.pop_back();
       if (tag.front() == '\'')
         tag.remove_prefix(1);
       push_many(ae->tag, tag.begin(), tag.size());
diff --git a/vowpalwabbit/parser.cc b/vowpalwabbit/parser.cc
index 1a2331986fe..0908fc4398f 100644
--- a/vowpalwabbit/parser.cc
+++ b/vowpalwabbit/parser.cc
@@ -141,7 +141,8 @@ void reset_source(vw& all, size_t numbits)
         input->close_file();
       else
       {
-        int fd = input->files.pop();
+        int fd = input->files.back();
+        input->files.pop_back();
         const auto& fps = all.final_prediction_sink;
 
         // If the current popped file is not in the list of final predictions sinks, close it.
@@ -208,7 +209,7 @@ void finalize_source(parser* p)
 #else
   int f = fileno(stdin);
 #endif
-  while (!p->input->files.empty() && p->input->files.last() == f) p->input->files.pop();
+  while (!p->input->files.empty() && p->input->files.back() == f) p->input->files.pop_back();
   p->input->close_files();
 
   delete p->input;
@@ -586,9 +587,9 @@ void set_done(vw& all)
 void addgrams(vw& all, size_t ngram, size_t skip_gram, features& fs, size_t initial_length, v_array<size_t>& gram_mask,
     size_t skips)
 {
-  if (ngram == 0 && gram_mask.last() < initial_length)
+  if (ngram == 0 && gram_mask.back() < initial_length)
   {
-    size_t last = initial_length - gram_mask.last();
+    size_t last = initial_length - gram_mask.back();
     for (size_t i = 0; i < last; i++)
     {
       uint64_t new_index = fs.indicies[i];
@@ -610,9 +611,9 @@ void addgrams(vw& all, size_t ngram, size_t skip_gram, features& fs, size_t init
   }
   if (ngram > 0)
   {
-    gram_mask.push_back(gram_mask.last() + 1 + skips);
+    gram_mask.push_back(gram_mask.back() + 1 + skips);
     addgrams(all, ngram - 1, skip_gram, fs, initial_length, gram_mask, 0);
-    gram_mask.pop();
+    gram_mask.pop_back();
   }
   if (skip_gram > 0 && ngram > 0)
     addgrams(all, ngram, skip_gram - 1, fs, initial_length, gram_mask, skips + 1);
diff --git a/vowpalwabbit/recall_tree.cc b/vowpalwabbit/recall_tree.cc
index 431ada1eea8..5eae74379a8 100644
--- a/vowpalwabbit/recall_tree.cc
+++ b/vowpalwabbit/recall_tree.cc
@@ -246,7 +246,7 @@ void remove_node_id_feature(recall_tree& /* b */, uint32_t /* cn */, example& ec
 {
   features& fs = ec.feature_space[node_id_namespace];
   fs.clear();
-  ec.indices.pop();
+  ec.indices.pop_back();
 }
 
 uint32_t oas_predict(recall_tree& b, single_learner& base, uint32_t cn, example& ec)
diff --git a/vowpalwabbit/search.cc b/vowpalwabbit/search.cc
index 5df4c56d98b..43c98d08395 100644
--- a/vowpalwabbit/search.cc
+++ b/vowpalwabbit/search.cc
@@ -617,7 +617,7 @@ void add_new_feature(search_private& priv, float val, uint64_t idx)
   uint64_t idx2 = ((idx & mask) >> ss) & mask;
   features& fs = priv.dat_new_feature_ec->feature_space[priv.dat_new_feature_namespace];
   fs.push_back(val * priv.dat_new_feature_value, ((priv.dat_new_feature_idx + idx2) << ss));
-  cdbg << "adding: " << fs.indicies.last() << ':' << fs.values.last() << endl;
+  cdbg << "adding: " << fs.indicies.back() << ':' << fs.values.back() << endl;
   if (priv.all->audit)
   {
     std::stringstream temp;
@@ -628,17 +628,17 @@ void add_new_feature(search_private& priv, float val, uint64_t idx)
 
 void del_features_in_top_namespace(search_private& /* priv */, example& ec, size_t ns)
 {
-  if ((ec.indices.size() == 0) || (ec.indices.last() != ns))
+  if ((ec.indices.size() == 0) || (ec.indices.back() != ns))
   {
     return;
     // if (ec.indices.size() == 0)
     //{ THROW("internal error (bug): expecting top namespace to be '" << ns << "' but it was empty"); }
     // else
     //{ THROW("internal error (bug): expecting top namespace to be '" << ns << "' but it was " <<
-    //(size_t)ec.indices.last()); }
+    //(size_t)ec.indices.back()); }
   }
   features& fs = ec.feature_space[ns];
-  ec.indices.decr();
+  ec.indices.pop_back();
   ec.num_features -= fs.size();
   ec.total_sum_feat_sq -= fs.sum_feat_sq;
   fs.clear();
@@ -895,7 +895,7 @@ void add_example_conditioning(search_private& priv, example& ec, size_t conditio
 
 void del_example_conditioning(search_private& priv, example& ec)
 {
-  if ((ec.indices.size() > 0) && (ec.indices.last() == conditioning_namespace))
+  if ((ec.indices.size() > 0) && (ec.indices.back() == conditioning_namespace))
     del_features_in_top_namespace(priv, ec, conditioning_namespace);
 }
 
@@ -1080,7 +1080,7 @@ void allowed_actions_to_label(search_private& priv, size_t ec_cnt, const action*
 template <class T>
 void ensure_size(v_array<T>& A, size_t sz)
 {
-  if ((size_t)(A.end_array - A.begin()) < sz)
+  if (A.capacity() < sz)
     A.resize(sz * 2 + 1);
   A.end() = A.begin() + sz;
 }
@@ -3484,7 +3484,7 @@ action predictor::predict()
       : sch.predict(*ec, my_tag, orA, oracle_actions.size(), cOn, cNa, alA, numAlA, alAcosts, learner_id, weight);
 
   if (condition_on_names.size() > 0)
-    condition_on_names.pop();  // un-null-terminate
+    condition_on_names.pop_back();  // un-null-terminate
   return p;
 }
 }  // namespace Search
diff --git a/vowpalwabbit/search_dep_parser.cc b/vowpalwabbit/search_dep_parser.cc
index d501100fb55..1976af51bea 100644
--- a/vowpalwabbit/search_dep_parser.cc
+++ b/vowpalwabbit/search_dep_parser.cc
@@ -162,7 +162,7 @@ size_t transition_hybrid(Search::search &sch, uint64_t a_id, uint32_t idx, uint3
   }
   else if (a_id == REDUCE_RIGHT)
   {
-    uint32_t last = stack.last();
+    uint32_t last = stack.back();
     uint32_t hd = stack[stack.size() - 2];
     heads[last] = hd;
     children[5][hd] = children[4][hd];
@@ -171,12 +171,12 @@ size_t transition_hybrid(Search::search &sch, uint64_t a_id, uint32_t idx, uint3
     tags[last] = t_id;
     sch.loss(gold_heads[last] != heads[last] ? 2 : (gold_tags[last] != t_id) ? 1.f : 0.f);
     assert(!stack.empty());
-    stack.pop();
+    stack.pop_back();
     return idx;
   }
   else if (a_id == REDUCE_LEFT)
   {
-    size_t last = stack.last();
+    size_t last = stack.back();
     uint32_t hd = idx;
     heads[last] = hd;
     children[3][hd] = children[2][hd];
@@ -185,7 +185,7 @@ size_t transition_hybrid(Search::search &sch, uint64_t a_id, uint32_t idx, uint3
     tags[last] = t_id;
     sch.loss(gold_heads[last] != heads[last] ? 2 : (gold_tags[last] != t_id) ? 1.f : 0.f);
     assert(!stack.empty());
-    stack.pop();
+    stack.pop_back();
     return idx;
   }
   THROW("transition_hybrid failed");
@@ -205,7 +205,7 @@ size_t transition_eager(Search::search &sch, uint64_t a_id, uint32_t idx, uint32
   }
   else if (a_id == REDUCE_RIGHT)
   {
-    uint32_t hd = stack.last();
+    uint32_t hd = stack.back();
     stack.push_back(idx);
     uint32_t last = idx;
     heads[last] = hd;
@@ -218,7 +218,7 @@ size_t transition_eager(Search::search &sch, uint64_t a_id, uint32_t idx, uint32
   }
   else if (a_id == REDUCE_LEFT)
   {
-    size_t last = stack.last();
+    size_t last = stack.back();
     uint32_t hd = (idx > n) ? 0 : idx;
     heads[last] = hd;
     children[3][hd] = children[2][hd];
@@ -227,13 +227,13 @@ size_t transition_eager(Search::search &sch, uint64_t a_id, uint32_t idx, uint32
     tags[last] = t_id;
     sch.loss(gold_heads[last] != heads[last] ? 2 : (gold_tags[last] != t_id) ? 1.f : 0.f);
     assert(!stack.empty());
-    stack.pop();
+    stack.pop_back();
     return idx;
   }
   else if (a_id == REDUCE)
   {
     assert(!stack.empty());
-    stack.pop();
+    stack.pop_back();
     return idx;
   }
   THROW("transition_eager failed");
@@ -253,7 +253,7 @@ void extract_features(Search::search &sch, uint32_t idx, multi_ex &ec)
 
   size_t n = ec.size();
   bool empty = stack.empty();
-  size_t last = empty ? 0 : stack.last();
+  size_t last = empty ? 0 : stack.back();
 
   for (size_t i = 0; i < 13; i++) ec_buf[i] = nullptr;
 
@@ -345,7 +345,7 @@ void get_valid_actions(Search::search &sch, v_array<uint32_t> &valid_action, uin
 
     if (stack_depth == 0)
       temp[REDUCE] = 0;
-    else if (idx <= n + 1 && heads[stack.last()] == my_null)
+    else if (idx <= n + 1 && heads[stack.back()] == my_null)
       temp[REDUCE] = 0;
 
     if (stack_depth == 0)
@@ -355,7 +355,7 @@ void get_valid_actions(Search::search &sch, v_array<uint32_t> &valid_action, uin
     }
     else
     {
-      if (heads[stack.last()] != my_null)
+      if (heads[stack.back()] != my_null)
         temp[REDUCE_LEFT] = 0;
       if (idx <= n && heads[idx] != my_null)
         temp[REDUCE_RIGHT] = 0;
@@ -382,7 +382,7 @@ void get_eager_action_cost(Search::search &sch, uint32_t idx, uint64_t n)
   v_array<uint32_t> &action_loss = data->action_loss, &stack = data->stack, &gold_heads = data->gold_heads,
                     heads = data->heads;
   size_t size = stack.size();
-  size_t last = (size == 0) ? 0 : stack.last();
+  size_t last = (size == 0) ? 0 : stack.back();
   for (size_t i = 1; i <= 4; i++) action_loss[i] = 0;
   if (!stack.empty())
     for (size_t i = 0; i < size; i++)
@@ -422,7 +422,7 @@ void get_hybrid_action_cost(Search::search &sch, size_t idx, uint64_t n)
   task_data *data = sch.get_task_data<task_data>();
   v_array<uint32_t> &action_loss = data->action_loss, &stack = data->stack, &gold_heads = data->gold_heads;
   size_t size = stack.size();
-  size_t last = (size == 0) ? 0 : stack.last();
+  size_t last = (size == 0) ? 0 : stack.back();
 
   for (size_t i = 1; i <= 3; i++) action_loss[i] = 0;
   if (!stack.empty())
@@ -491,7 +491,7 @@ void get_gold_actions(Search::search &sch, uint32_t idx, uint64_t /* n */, v_arr
                     &valid_actions = data->valid_actions;
   gold_actions.clear();
   size_t size = stack.size();
-  size_t last = (size == 0) ? 0 : stack.last();
+  size_t last = (size == 0) ? 0 : stack.back();
   uint32_t &sys = data->transition_system;
 
   if (sys == arc_hybrid && is_valid(SHIFT, valid_actions) && (stack.empty() || gold_heads[idx] == last))
@@ -627,16 +627,16 @@ void run(Search::search &sch, multi_ex &ec)
       extract_features(sch, idx, ec);
       computedFeatures = true;
     }
-    get_valid_actions(sch, valid_actions, idx, n, (uint64_t)stack.size(), stack.empty() ? 0 : stack.last());
+    get_valid_actions(sch, valid_actions, idx, n, (uint64_t)stack.size(), stack.empty() ? 0 : stack.back());
     if (sys == arc_hybrid)
       get_hybrid_action_cost(sch, idx, n);
     else if (sys == arc_eager)
       get_eager_action_cost(sch, idx, n);
 
     // get gold tag labels
-    left_label = stack.empty() ? my_null : gold_tags[stack.last()];
+    left_label = stack.empty() ? my_null : gold_tags[stack.back()];
     if (sys == arc_hybrid)
-      right_label = stack.empty() ? my_null : gold_tags[stack.last()];
+      right_label = stack.empty() ? my_null : gold_tags[stack.back()];
     else if (sys == arc_eager)
       right_label = idx <= n ? gold_tags[idx] : (uint32_t)data->root_label;
     else
@@ -751,9 +751,9 @@ void run(Search::search &sch, multi_ex &ec)
   }
   if (sys == arc_hybrid)
   {
-    heads[stack.last()] = 0;
-    tags[stack.last()] = (uint32_t)data->root_label;
-    sch.loss((gold_heads[stack.last()] != heads[stack.last()]));
+    heads[stack.back()] = 0;
+    tags[stack.back()] = (uint32_t)data->root_label;
+    sch.loss((gold_heads[stack.back()] != heads[stack.back()]));
   }
   if (sch.output().good())
     for (size_t i = 1; i <= n; i++) sch.output() << (heads[i]) << ":" << tags[i] << std::endl;
diff --git a/vowpalwabbit/search_entityrelationtask.cc b/vowpalwabbit/search_entityrelationtask.cc
index 0ce16573a36..8e17f3f79ca 100644
--- a/vowpalwabbit/search_entityrelationtask.cc
+++ b/vowpalwabbit/search_entityrelationtask.cc
@@ -155,7 +155,7 @@ size_t predict_entity(
                      .set_allowed(my_task_data->y_allowed_entity)
                      .set_learner_id(1)
                      .predict();
-    my_task_data->y_allowed_entity.pop();
+    my_task_data->y_allowed_entity.pop_back();
   }
   else
   {
@@ -240,7 +240,7 @@ size_t predict_relation(Search::search& sch, example* ex, v_array<size_t>& predi
                      .add_condition(id1, 'a')
                      .add_condition(id2, 'b')
                      .predict();
-    constrained_relation_labels.pop();
+    constrained_relation_labels.pop_back();
   }
   else
   {
diff --git a/vowpalwabbit/search_graph.cc b/vowpalwabbit/search_graph.cc
index dc7d7a9f546..1e557316cc2 100644
--- a/vowpalwabbit/search_graph.cc
+++ b/vowpalwabbit/search_graph.cc
@@ -354,7 +354,7 @@ void add_edge_features(Search::search& sch, task_data& D, size_t n, multi_ex& ec
 
 void del_edge_features(task_data& /*D*/, uint32_t n, multi_ex& ec)
 {
-  ec[n]->indices.pop();
+  ec[n]->indices.pop_back();
   features& fs = ec[n]->feature_space[neighbor_namespace];
   ec[n]->total_sum_feat_sq -= fs.sum_feat_sq;
   ec[n]->num_features -= fs.size();
diff --git a/vowpalwabbit/slim/src/vw_slim_predict.cc b/vowpalwabbit/slim/src/vw_slim_predict.cc
index 66c79274e3e..4b3e154f2f2 100644
--- a/vowpalwabbit/slim/src/vw_slim_predict.cc
+++ b/vowpalwabbit/slim/src/vw_slim_predict.cc
@@ -26,7 +26,7 @@ namespace_copy_guard::namespace_copy_guard(example_predict& ex, unsigned char ns
 
 namespace_copy_guard::~namespace_copy_guard()
 {
-  _ex.indices.pop();
+  _ex.indices.pop_back();
   if (_remove_ns)
     _ex.feature_space[_ns].clear();
 }
@@ -58,4 +58,4 @@ stride_shift_guard::~stride_shift_guard()
       for (auto& f : _ex.feature_space[ns]) f.index() >>= _shift;
 }
 
-};  // namespace vw_slim
\ No newline at end of file
+};  // namespace vw_slim
diff --git a/vowpalwabbit/v_array.h b/vowpalwabbit/v_array.h
index 48b4ec9a72b..4695bd11350 100644
--- a/vowpalwabbit/v_array.h
+++ b/vowpalwabbit/v_array.h
@@ -21,6 +21,7 @@
 #include "vw_exception.h"
 #endif
 
+#include "future_compat.h"
 #include "memory.h"
 
 const size_t erase_point = ~((1u << 10u) - 1u);
@@ -52,21 +53,38 @@ struct v_array
   // ~v_array() {
   //  delete_v();
   // }
-  T last() const { return *(_end - 1); }
-  T pop() { return *(--_end); }
-  bool empty() const { return _begin == _end; }
-  void decr() { _end--; }
+  inline T& back() { return *(_end - 1); }
+  inline const T& back() const { return *(_end - 1); }
+  inline void pop_back() { (--_end)->~T(); }
+
+  VW_DEPRECATED("v_array::last() is deprecated. Use back()")
+  T last() const { return back(); }
+  
+  VW_DEPRECATED("v_array::pop() is deprecated. Use pop_back()")
+  T pop()
+  {
+    T ret = back();
+    pop_back();
+    return ret;
+  }
+  
+  inline bool empty() const { return _begin == _end; }
+
+  VW_DEPRECATED("v_array::decr() is deprecated. Use pop_back()")
+  inline void decr() { _end--; }
+  VW_DEPRECATED("v_array::incr() is deprecated.")
   void incr()
   {
     if (_end == end_array)
-      resize(2 * (end_array - _begin) + 3);
+      resize(2 * capacity() + 3);
     _end++;
   }
   T& operator[](size_t i) const { return _begin[i]; }
   inline size_t size() const { return _end - _begin; }
+  inline size_t capacity() const { return end_array - _begin; }
   void resize(size_t length)
   {
-    if ((size_t)(end_array - _begin) != length)
+    if (capacity() != length)
     {
       size_t old_len = _end - _begin;
       T* temp = (T*)realloc(_begin, sizeof(T) * length);
@@ -83,6 +101,27 @@ struct v_array
     }
   }
 
+  // insert() and remove() don't follow the standard spec, which calls for iterators
+  // instead of indices. But these fn signatures follow our usage better.
+  // These functions do not check bounds, undefined behavior if they are called
+  // on out-of-bounds indices
+  // insert before the indexed element
+  inline void insert(size_t idx, const T& elem)
+  {
+    if (_end == end_array)
+      resize(2 * capacity() + 3);
+    _end++;
+    memmove(&_begin[idx + 1], &_begin[idx], (size() - (idx + 1)) * sizeof(T));
+    _begin[idx] = elem;
+  }
+  // erase indexed element
+  inline void erase(size_t idx)
+  {
+    _begin[idx].~T();
+    memmove(&_begin[idx], &_begin[idx + 1], (size() - (idx + 1)) * sizeof(T));
+    --_end;
+  }
+
   void clear()
   {
     if (++erase_count & erase_point)
@@ -105,7 +144,7 @@ struct v_array
   void push_back(const T& new_ele)
   {
     if (_end == end_array)
-      resize(2 * (end_array - _begin) + 3);
+      resize(2 * capacity() + 3);
     new (_end++) T(new_ele);
   }
 
@@ -115,7 +154,7 @@ struct v_array
   void emplace_back(Args&&... args)
   {
     if (_end == end_array)
-      resize(2 * (end_array - _begin) + 3);
+      resize(2 * capacity() + 3);
     new (_end++) T(std::forward<Args>(args)...);
   }
 
@@ -153,7 +192,7 @@ struct v_array
     if (!contain_sorted(new_ele, index))
     {
       if (_end == end_array)
-        resize(2 * (end_array - _begin) + 3);
+        resize(2 * capacity() + 3);
 
       to_move = size - index;
 
@@ -214,7 +253,7 @@ template <class T>
 void push_many(v_array<T>& v, const T* _begin, size_t num)
 {
   if (v._end + num >= v.end_array)
-    v.resize(std::max(2 * (size_t)(v.end_array - v._begin) + 3, v._end - v._begin + num));
+    v.resize(std::max(2 * v.capacity() + 3, v.size() + num));
 #ifdef _WIN32
   memcpy_s(v._end, v.size() - (num * sizeof(T)), _begin, num * sizeof(T));
 #else
diff --git a/vowpalwabbit/vwdll.cpp b/vowpalwabbit/vwdll.cpp
index 7bb806c8346..71f09c8f48c 100644
--- a/vowpalwabbit/vwdll.cpp
+++ b/vowpalwabbit/vwdll.cpp
@@ -368,7 +368,7 @@ class memory_io_buf : public io_buf
 
     virtual bool close_file() {
        if (files.size() > 0) {
-            files.pop();
+            files.pop_back();
             return true;
         }
         return false;

From 2fbc73fa5719c0b1d9c3a557e58849a4b0abaa2f Mon Sep 17 00:00:00 2001
From: peterychang <petchang.microsoft.com>
Date: Thu, 2 Jan 2020 17:06:27 -0500
Subject: [PATCH 2/3] add r-value versions of functions

---
 vowpalwabbit/v_array.h | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/vowpalwabbit/v_array.h b/vowpalwabbit/v_array.h
index 4695bd11350..2b157dea32b 100644
--- a/vowpalwabbit/v_array.h
+++ b/vowpalwabbit/v_array.h
@@ -110,9 +110,17 @@ struct v_array
   {
     if (_end == end_array)
       resize(2 * capacity() + 3);
-    _end++;
+    ++_end;
+    memmove(&_begin[idx + 1], &_begin[idx], (size() - (idx + 1)) * sizeof(T));
+    new (&_begin[idx]) T(elem);
+  }
+  inline void insert(size_t idx, T&& elem)
+  {
+    if (_end == end_array)
+      resize(2 * capacity() + 3);
+    ++_end;
     memmove(&_begin[idx + 1], &_begin[idx], (size() - (idx + 1)) * sizeof(T));
-    _begin[idx] = elem;
+    new (&_begin[idx]) T(std::move(elem));
   }
   // erase indexed element
   inline void erase(size_t idx)
@@ -147,6 +155,12 @@ struct v_array
       resize(2 * capacity() + 3);
     new (_end++) T(new_ele);
   }
+  void push_back(T&& new_ele)
+  {
+    if (_end == end_array)
+      resize(2 * capacity() + 3);
+    new (_end++) T(std::move(new_ele));
+  }
 
   void push_back_unchecked(const T& new_ele) { new (_end++) T(new_ele); }
 

From 792d83233d0e1bcfde9030e90ab51003c37951b5 Mon Sep 17 00:00:00 2001
From: peterychang <petchang.microsoft.com>
Date: Fri, 3 Jan 2020 11:56:47 -0500
Subject: [PATCH 3/3] add asserts and comments

---
 vowpalwabbit/v_array.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/vowpalwabbit/v_array.h b/vowpalwabbit/v_array.h
index 2b157dea32b..b2fe504f870 100644
--- a/vowpalwabbit/v_array.h
+++ b/vowpalwabbit/v_array.h
@@ -104,10 +104,14 @@ struct v_array
   // insert() and remove() don't follow the standard spec, which calls for iterators
   // instead of indices. But these fn signatures follow our usage better.
   // These functions do not check bounds, undefined behavior if they are called
-  // on out-of-bounds indices
+  // on out-of-bounds indices.
+  // NOTE: this function is only safe for trivially copyable objects
+  // any object that contains internal pointers or refernces will break
+  // when using these functions
   // insert before the indexed element
   inline void insert(size_t idx, const T& elem)
   {
+    assert(idx <= size());
     if (_end == end_array)
       resize(2 * capacity() + 3);
     ++_end;
@@ -116,6 +120,7 @@ struct v_array
   }
   inline void insert(size_t idx, T&& elem)
   {
+    assert(idx <= size());
     if (_end == end_array)
       resize(2 * capacity() + 3);
     ++_end;
@@ -125,6 +130,7 @@ struct v_array
   // erase indexed element
   inline void erase(size_t idx)
   {
+    assert(idx < size());
     _begin[idx].~T();
     memmove(&_begin[idx], &_begin[idx + 1], (size() - (idx + 1)) * sizeof(T));
     --_end;