szaszm commented on a change in pull request #1096:
URL: https://github.com/apache/nifi-minifi-cpp/pull/1096#discussion_r644725588
##########
File path: extensions/tensorflow/CMakeLists.txt
##########
@@ -26,7 +26,7 @@ find_package(TensorFlow REQUIRED)
message("-- Found TensorFlow: ${TENSORFLOW_INCLUDE_DIRS}")
-include_directories(${TENSORFLOW_INCLUDE_DIRS})
+include_directories(SYSTEM ${TENSORFLOW_INCLUDE_DIRS})
Review comment:
Can we change this to `target_include_directories` to the
`minifi-tensorflow-extensions` target? Preferably PRIVATE.
##########
File path: extensions/tensorflow/TFApplyGraph.cpp
##########
@@ -194,7 +197,7 @@ int64_t TFApplyGraph::GraphReadCallback::process(const
std::shared_ptr<io::BaseS
auto num_read = stream->read(reinterpret_cast<uint8_t
*>(&graph_proto_buf[0]),
static_cast<int>(stream->size()));
- if (num_read != stream->size()) {
+ if (static_cast<uint64_t>(num_read) != stream->size()) {
Review comment:
`stream->size()` returns `size_t`, not `uint64_t`
##########
File path: extensions/tensorflow/TFExtractTopLabels.cpp
##########
@@ -133,8 +136,11 @@ int64_t
TFExtractTopLabels::LabelsReadCallback::process(const std::shared_ptr<io
while (total_read < stream->size()) {
auto read = stream->read(reinterpret_cast<uint8_t *>(&buf[0]),
static_cast<int>(buf_size));
+ if (read < 0) {
Review comment:
This will need to be changed after my PR #1028 is merged, but it might
not conflict, because this is new code. If this slipped through, it would mean
that `read` is never smaller than zero and line 143 becomes a loop to the
largest `size_t` value.
Could you specify the type on line 138 to be int and use brace
initialization (direct-list-initialization) to make this an error when
`stream->read()` returns a `size_t`? Or add a static_assert that the variable
is indeed of type int.
##########
File path: extensions/tensorflow/TFConvertImageToTensor.cpp
##########
@@ -325,7 +325,7 @@ int64_t
TFConvertImageToTensor::ImageReadCallback::process(const std::shared_ptr
auto num_read = stream->read(tensor_->flat<unsigned char>().data(),
static_cast<int>(stream->size()));
- if (num_read != stream->size()) {
+ if (static_cast<uint64_t>(num_read) != stream->size()) {
Review comment:
same here: `stream->size()` returns `size_t`, so we should preferably
cast `num_read` to that. Same at line 340.
##########
File path: extensions/tensorflow/TFApplyGraph.cpp
##########
@@ -221,7 +224,7 @@ int64_t TFApplyGraph::TensorWriteCallback::process(const
std::shared_ptr<io::Bas
auto num_wrote = stream->write(reinterpret_cast<uint8_t
*>(&tensor_proto_buf[0]),
static_cast<int>(tensor_proto_buf.size()));
- if (num_wrote != tensor_proto_buf.size()) {
+ if (static_cast<uint64_t>(num_wrote) != tensor_proto_buf.size()) {
Review comment:
same here: `stream->size()` returns `size_t`, so we should preferably
cast `num_wrote` to that.
##########
File path: extensions/tensorflow/TFExtractTopLabels.cpp
##########
@@ -156,7 +162,7 @@ int64_t
TFExtractTopLabels::TensorReadCallback::process(const std::shared_ptr<io
auto num_read = stream->read(reinterpret_cast<uint8_t
*>(&tensor_proto_buf[0]),
static_cast<int>(stream->size()));
- if (num_read != stream->size()) {
+ if (static_cast<uint64_t>(num_read) != stream->size()) {
Review comment:
same here: `stream->size()` returns `size_t`, so we should preferably
cast `num_read` to that.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
