Merge pull request torch#256 from colesbury/lua52

Rename unpack to table.unpack for Lua 5.2

CriterionTable.lua

@@ -7,11 +7,11 @@ function CriterionTable:__init(criterion)
 end
 
 function CriterionTable:updateOutput(input) 
-   self.output = self.criterion:updateOutput(unpack(input))
+   self.output = self.criterion:updateOutput(table.unpack(input))
    return self.output
 end
 
 function CriterionTable:updateGradInput(input, gradOutput)
-  self.criterion:updateGradInput(unpack(input))
+  self.criterion:updateGradInput(table.unpack(input))
   return self.gradInput
 end 

Euclidean.lua

@@ -35,9 +35,9 @@ end
 local function view(res, src, ...)
    local args = {...}
    if src:isContiguous() then
-      res:view(src, unpack(args))
+      res:view(src, table.unpack(args))
    else
-      res:reshape(src, unpack(args))
+      res:reshape(src, table.unpack(args))
    end
 end
 

MM.lua

@@ -18,7 +18,7 @@ end
 
 function MM:updateOutput(input)
   assert(#input == 2, 'input must be a pair of minibatch matrices')
-  local a, b = unpack(input)
+  local a, b = table.unpack(input)
   assert(a:nDimension() == 2 or a:nDimension() == 3, 'input tensors must be 2D or 3D')
 
   if a:nDimension() == 2 then
@@ -47,7 +47,7 @@ end
 
 function MM:updateGradInput(input, gradOutput)
   assert(#input == 2, 'input must be a pair of tensors')
-  local a, b = unpack(input)
+  local a, b = table.unpack(input)
   self.gradInput[1]:resizeAs(a)
   self.gradInput[2]:resizeAs(b)
 

MixtureTable.lua

@@ -18,7 +18,7 @@ function MixtureTable:__init(dim)
 end
 
 function MixtureTable:updateOutput(input) 
-   local gaterInput, expertInputs = unpack(input)
+   local gaterInput, expertInputs = table.unpack(input)
 
    self.dimG = 2
    local batchSize = gaterInput:size(1)
@@ -79,8 +79,8 @@ function MixtureTable:updateOutput(input)
 end
 
 function MixtureTable:updateGradInput(input, gradOutput)
-   local gaterInput, expertInputs = unpack(input)
-   local gaterGradInput, expertGradInputs = unpack(self.gradInput)
+   local gaterInput, expertInputs = table.unpack(input)
+   local gaterGradInput, expertGradInputs = table.unpack(self.gradInput)
 
    if self.table then
       if not self.backwardSetup then

Module.lua

@@ -146,7 +146,7 @@ function Module:getParameters()
       if storageAndOffset == nil then
           return nil
       end
-      local _, offset = unpack(storageAndOffset)
+      local _, offset = table.unpack(storageAndOffset)
       return offset
    end
 
@@ -200,7 +200,7 @@ function Module:getParameters()
       end
 
       for _, storageAndOffset in pairs(storages) do
-         local k, v = unpack(storageAndOffset)
+         local k, v = table.unpack(storageAndOffset)
          flatParameters[{{v+1,v+k:size()}}]:copy(Tensor():set(k))
       end
 

View.lua

@@ -74,7 +74,7 @@ end
 function View:updateOutput(input)
    local bsz = batchsize(input, self.size, self.numInputDims, self.numElements)
    if bsz then
-      self.output = input:view(bsz, unpack(self.size:totable()))
+      self.output = input:view(bsz, table.unpack(self.size:totable()))
    else
       self.output = input:view(self.size)
    end

WeightedEuclidean.lua

@@ -26,9 +26,9 @@ end
 local function view(res, src, ...)
    local args = {...}
    if src:isContiguous() then
-      res:view(src, unpack(args))
+      res:view(src, table.unpack(args))
    else
-      res:reshape(src, unpack(args))
+      res:reshape(src, table.unpack(args))
    end
 end
 

doc/table.md

@@ -610,7 +610,7 @@ Example 1:
 -0.2955
 [torch.DoubleTensor of dimension 2x1]
 
-> =unpack(nn.SelectTable(1):backward(input, torch.randn(2, 3)))
+> =table.unpack(nn.SelectTable(1):backward(input, torch.randn(2, 3)))
 -0.4891 -0.3495 -0.3182
 -2.0999  0.7381 -0.5312
 [torch.DoubleTensor of dimension 2x3]
@@ -634,7 +634,7 @@ Example 2:
     }
 }
 
-> =unpack(nn.SelectTable(2):backward(input, {torch.randn(2, 1), {torch.randn(2, 2)}}))
+> =table.unpack(nn.SelectTable(2):backward(input, {torch.randn(2, 1), {torch.randn(2, 2)}}))
 0 0 0
 0 0 0
 [torch.DoubleTensor of dimension 2x3]

hessian.lua

@@ -330,7 +330,7 @@ function nn.hessian.enable()
          if storageAndOffset == nil then
              return nil
          end
-         local _, offset = unpack(storageAndOffset)
+         local _, offset = table.unpack(storageAndOffset)
          return offset
       end
 
@@ -373,7 +373,7 @@ function nn.hessian.enable()
          end
 
          for _, storageAndOffset in pairs(storages) do
-            local k, v = unpack(storageAndOffset)
+            local k, v = table.unpack(storageAndOffset)
             flatParameters[{{v+1,v+k:size()}}]:copy(torch.Tensor():set(k))
          end
          for k = 1,flatUsedParameters:nElement() do

test.lua

@@ -2923,7 +2923,7 @@ function nntest.View()
    local target = template:size():totable()
    local module = nn.View(template:size())
    mytester:assertTableEq(module:forward(input):size():totable(), target, "Error in forward (1)")
-   local module = nn.View(unpack(target))
+   local module = nn.View(table.unpack(target))
    mytester:assertTableEq(module:forward(input):size():totable(), target, "Error in forward (2)")
 
    -- Minibatch
@@ -2979,7 +2979,7 @@ function nntest.Reshape()
    local target = template:size():totable()
    local module = nn.Reshape(template:size())
    mytester:assertTableEq(module:forward(input):size():totable(), target, "Error in forward (1)")
-   local module = nn.View(unpack(target))
+   local module = nn.View(table.unpack(target))
    mytester:assertTableEq(module:forward(input):size():totable(), target, "Error in forward (2)")
 
    -- Minibatch
@@ -3005,7 +3005,7 @@ function nntest.SpatialUpSamplingNearest()
     end
 
     -- Check that the gradient is correct by using finite elements
-    local input = torch.Tensor(unpack(shape)):zero()
+    local input = torch.Tensor(table.unpack(shape)):zero()
 
     local err = jac.testJacobian(m, input)
     mytester:assertlt(err, precision, ' error on state ')
@@ -3250,7 +3250,7 @@ function nntest.MM()
   local gradOutput = torch.randn(M, P)
   local gradInput = mm:backward({A, B}, gradOutput)
   mytester:assert(#gradInput == 2, 'gradInput must be table of size 2')
-  local gradA, gradB = unpack(gradInput)
+  local gradA, gradB = table.unpack(gradInput)
   mytester:assertTableEq(gradA:size():totable(), A:size():totable(),
                          'Gradient for input A has wrong size')
   mytester:assertTableEq(gradB:size():totable(), B:size():totable(),
@@ -3281,7 +3281,7 @@ function nntest.BatchMMNoTranspose()
     local gradOutput = torch.randn(bSize, M, P)
     local gradInput = mm:backward({A, B}, gradOutput)
     mytester:assert(#gradInput == 2, 'gradInput must be table of size 2')
-    local gradA, gradB = unpack(gradInput)
+    local gradA, gradB = table.unpack(gradInput)
     mytester:assertTableEq(gradA:size():totable(), A:size():totable(),
                            'Gradient for input A has wrong size')
     mytester:assertTableEq(gradB:size():totable(), B:size():totable(),
@@ -3315,7 +3315,7 @@ function nntest.BatchMMTransposeA()
     local gradOutput = torch.randn(bSize, M, P)
     local gradInput = mm:backward({A, B}, gradOutput)
     mytester:assert(#gradInput == 2, 'gradInput must be table of size 2')
-    local gradA, gradB = unpack(gradInput)
+    local gradA, gradB = table.unpack(gradInput)
     mytester:assertTableEq(gradA:size():totable(), A:size():totable(),
                            'Gradient for input A has wrong size')
     mytester:assertTableEq(gradB:size():totable(), B:size():totable(),
@@ -3349,7 +3349,7 @@ function nntest.BatchMMTransposeB()
     local gradOutput = torch.randn(bSize, M, P)
     local gradInput = mm:backward({A, B}, gradOutput)
     mytester:assert(#gradInput == 2, 'gradInput must be table of size 2')
-    local gradA, gradB = unpack(gradInput)
+    local gradA, gradB = table.unpack(gradInput)
     mytester:assertTableEq(gradA:size():totable(), A:size():totable(),
                            'Gradient for input A has wrong size')
     mytester:assertTableEq(gradB:size():totable(), B:size():totable(),
@@ -3383,7 +3383,7 @@ function nntest.BatchMMTransposeBoth()
     local gradOutput = torch.randn(bSize, M, P)
     local gradInput = mm:backward({A, B}, gradOutput)
     mytester:assert(#gradInput == 2, 'gradInput must be table of size 2')
-    local gradA, gradB = unpack(gradInput)
+    local gradA, gradB = table.unpack(gradInput)
     mytester:assertTableEq(gradA:size():totable(), A:size():totable(),
                            'Gradient for input A has wrong size')
     mytester:assertTableEq(gradB:size():totable(), B:size():totable(),