In Xcode has a build error:Array input is not a constant array of tensors

[TuomaXu]

code:
import TensorFlow
let x = Tensor([[1, 2], [3, 4]]) //error：Array input is not a constant array of tensors
print(x)

i have install tensorflow-swift。Xcode Version 9.3 (9E145) and Selected a Swift toolchain affects the Xcode。

but the same code is ok in Swift REPL

[rxwei]

Have you enabled the optimizer by passing the -O flag to the compiler?

[TuomaXu]

@rxwei i do not . and how to passing the -O flag to the compiler? thanks!

[frogcjn]

@TuomaXu Choose Build Settings -> Swift Compiler!!!! -> Code Generation -> Optimization Level [-O]

@rxwei , after I choose to Optimize for Speed, I still get error:

"Undefined symbols for architecture x86_64" error appears at Link step:

Link ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Products/Debug/TensorFlowTest

Ld ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Products/Debug/TensorFlowTest normal x86_64
    cd ~/Desktop/TensorFlowTest
    export MACOSX_DEPLOYMENT_TARGET=10.13
    /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang -arch x86_64 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.13.sdk -L~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Products/Debug -F~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Products/Debug -filelist ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Intermediates.noindex/TensorFlowTest.build/Debug/TensorFlowTest.build/Objects-normal/x86_64/TensorFlowTest.LinkFileList -mmacosx-version-min=10.13 -Xlinker -object_path_lto -Xlinker ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Intermediates.noindex/TensorFlowTest.build/Debug/TensorFlowTest.build/Objects-normal/x86_64/TensorFlowTest_lto.o -Xlinker -export_dynamic -Xlinker -no_deduplicate -fobjc-link-runtime -L/Library/Developer/Toolchains/swift-tensorflow-DEVELOPMENT-2018-04-26-a.xctoolchain/usr/lib/swift_static/macosx -Xlinker -force_load_swift_libs -lc++ -framework Foundation -Xlinker -add_ast_path -Xlinker ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Intermediates.noindex/TensorFlowTest.build/Debug/TensorFlowTest.build/Objects-normal/x86_64/TensorFlowTest.swiftmodule -Xlinker -dependency_info -Xlinker ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Intermediates.noindex/TensorFlowTest.build/Debug/TensorFlowTest.build/Objects-normal/x86_64/TensorFlowTest_dependency_info.dat -o ~/Library/Developer/Xcode/DerivedData/TensorFlowTest-dmfssjehvgbjmzfmfkuplflenfzq/Build/Products/Debug/TensorFlowTest

Undefined symbols for architecture x86_64:
  "_TFE_ContextAddFunction", referenced from:
      _$S10TensorFlow8TFEState33_F1EB29DDC3964F4495E8056659BC5464LLC_16programByteCount24entryFunctionNameAddressADSV_SiSPys4Int8VGtcfc in libswiftTensorFlow.a(TensorFlow.o)
  "_TFE_ContextListDevices", referenced from:
      _$S10TensorFlow17_ExecutionContextCACycfc in libswiftTensorFlow.a(TensorFlow.o)
      ...
  "tensorflow::port::InitMain(char const*, int*, char***)", referenced from:
      _InitTensorFlowRuntime in libswiftTensorFlow.a(ctensorflow_init.cpp.o)
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)

[rxwei]

Did you download and install the toolchain, or just build from source?

[Mamonaku]

Same problem, on linux Ubuntu, with a small difference:

this works:

swift -I/<...>/tfswift/usr/lib/swift/clang/include
Welcome to Swift version 4.2-dev (LLVM 04bdb56f3d, Clang b44dbbdf44). Type :help for assistance.
  1> import TensorFlow
  2> let x = Tensor([[1, 2], [3, 4]])
2018-04-28 18:22:43.037014: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX
x: TensorFlow.Tensor<Double> = [[1.0, 2.0], [3.0, 4.0]]
  3> print(x)
[[1.0, 2.0], [3.0, 4.0]]
  4>  

however with main.swift

let x = Tensor([[1, 2], [3, 4]])
print(x)

This works....

swift -O Sources/tf1/main.swift
[[1.0, 2.0], [3.0, 4.0]]

but this doesn't

swift build
'tf1' /<.... >/swiftTF/tf1: warning: Ignoring duplicate product 'tf1' 
Compile Swift Module 'tf1' (1 sources)
/< ... >/swiftTF/tf1/Sources/tf1/main.swift:7:16: error: array input is not a constant array of tensors
let x = Tensor([[1, 2], [3, 4]])
               ^~~~~~~~~~~~~~~
error: terminated(1): /<>/tfswift/usr/bin/swift-build-tool -f /<>/swiftTF/tf1/.build/debug.yaml main output:

Any Ideas?

[dan-zheng]

@Mamonaku Try swift build -Xswiftc -O. You still need to pass -O to run optimizations.

[Mamonaku]

Bingo!! thanks!

[frogcjn]

@rxwei I installed from toolchain.
The Xcode playground will work fine, but the Xcode Project will not.

@rxwei, @TuomaXu I solved the problem by manually adding tensorflow framework.

Step 1: Set -O

Build Settings -> Swift Compiler!!!! -> Code Generation -> Optimization Level

Optimize for Speed [-O]

Step 2: Add two *.so Library

General -> Linked Frameworks and Libraries
libtensorflow.so and libtensorflow_framework.so in folder

/Library/Developer/Toolchains/swift-tensorflow-DEVELOPMENT-2018-04-26-a.xctoolchain/usr/lib/swift/macosx/

Step 3: Set Runpath

Build Settings -> Linking -> Runpath Search Path:

/Library/Developer/Toolchains/swift-tensorflow-DEVELOPMENT-2018-04-26-a.xctoolchain/usr/lib/swift/macosx

[johndpope]

Note the paths above will be broken with each new release

You can simply use swift-latest which is symlink to latest toolchain.

/Library/Developer/Toolchains/swift-tensorflow-DEVELOPMENT-2018-04-26-a.xctoolchain/usr/lib/swift/macosx

->

/Library/Developer/Toolchains/swift-latest/usr/lib/swift/macosx

[johndpope]

It’s complaining it cant find symbols for python. Try adding the Xcode python2.7.tdb file to the frameworks

…

On Friday, September 21, 2018, Gosicfly ***@***.***> wrote: Cannot work well after setting as your workaround. Please help. @frogcjn <https://github.com/frogcjn> [image: image] <https://user-images.githubusercontent.com/11231087/45875281-beaec100-bdc9-11e8-885e-4162f2981d69.png> — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#10 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AARsyq3fcgHXJyws_mGnENkIVtn2FWAMks5udLtggaJpZM4TrTPZ> .