Co-authored-by: Akihito Nakano <sora.akatsuki@gmail.com>
Co-authored-by: Jeremie Bresson <dev@jmini.fr>
Co-authored-by: Jim Schubert <james.schubert@gmail.com>
Co-authored-by: Martin Delille <martin@phonations.com>
Co-authored-by: Tomasz Prus <tomasz.prus@gmail.com>
Co-authored-by: William Cheng <wing328hk@gmail.com>
* Small fixes to prevent crash when empty json body is provided.
* Fix some more memory Leaks in the model-body
- Members not deleted in cleanup() method, for maps/arrays of primitive types.
- Avoid undefined behavior when updating class members with data from missing json fields
* Small fixes to prevent crash when empty json body is provided.
* WIP: plug mem-leaks
* fixup: add the QJsonObject instance in toJsonArray instead of pointer
* fixup: simplify toJsonMap
Actually the original solution is incomplete, because "innerType" maps to a single C++ type. Have a look at Qt's builtin QJsonObject::fromVariantMap.
* Updates to antis81:patch-1 after tests.
* update to remove string allocation
* Updates due to address of members being passed
* Update PetStore Examples
* Small updates for Header includes
* Disable creation of empty json fields and fields for primitives which were not set, but using default values
modelnamePrefix will be the one passed from command line or SWG if none
* Updates after review
Also common http files are splitted
Update Petstore examples
* Small Fixes for cleaning up arrays of arrays/maps
* Changes
- Api Body pass by Reference, avoid seg fault due to deletion of auto constructed object passed as parameter.
- Support Maps and Arrays upto a depth of 2
- Refactor Helpers to handle Maps and simplify code
* Remove size check due to possible incorrect type
* Update PetStore Examples for Qt5 C++ and small fixes for QDate/QDateTime
* Updates after review. Fixes typo and remove usage of auto keyword.
* Restored usage of auto keyword.
* Disable creation of empty json fields and fields for primitives which were not set, but using default values
modelnamePrefix will be the one passed from command line or SWG if none
* Updates after review
Also common http files are splitted
Update Petstore examples
* - check C++ keywords
- setter for SSL configuration
* recreate example QT5CPP
* only emit error signal if error happens
* provide more error information
* regenerate example QT5CPP
* * use reserved words in abstract cpp generator
* update CppRest examples
* remove tabs
* update pom to prepare 2.2.3 release
* add header support to qt5, update samples
* update version in java help
* update to xcode 9
* rollback to xcode8.2
* Changing QBuffer to use a QByteArray solves the issue for me since there is no real use-case for using a QBuffer.
Documentation of QT5 states:
QBuffer::QBuffer(QByteArray *byteArray, QObject *parent = Q_NULLPTR)
Constructs a QBuffer that uses the QByteArray pointed to by byteArray as its internal buffer, and with the given parent. The caller is responsible for ensuring that byteArray remains valid until the QBuffer is destroyed, or until setBuffer() is called to change the buffer. QBuffer doesn't take ownership of the QByteArray.
Since the variable “request_content” is allocated on the stack, this is clearly wrong and a bug. The construction of QBuffer is designed this way so that whenever you write to the buffer, it is also written to the byte array that it is pointing to
* Add a retro-compatible solution based on QNetworkAccessManager SourceCode
* update samples
* Update samples for several languages.
Just whitespace changes and some reordering where the order doesn't matter.
* Adding generated .swagger-codegen/VERSION files.
* Add overloads on signals.
The overload adds the NetworkError and the error string to the signal. Both signal will be emitted. Clients are supposed to subscribe to only one of the overloads.
* refactor generated code alignement
* Do not use method overload for signals. Instead, have another naming convention for signal that carries Error information.
Why ? Because signal overloading and QT5 connector with method referencing do not mix well (see http://stackoverflow.com/questions/16794695/connecting-overloaded-signals-and-slots-in-qt-5). It would need to specify exactly which overload to take. That is cumbersome to write AND is a breaking change for the consumer of the generated code.
* update qt5 sample files
* Add support to UUID type
* Simplify and use a simple QString instead of a QUuid.
Work for using a QUuid is way larger than supporting QString (imply modifying multiple c++ templates).
Current needs are fullfill with the use of a simple QString, so I let the use of QUuid for future enhancement.
* - add windows executable for generating qt5 security samples
- update sample clients
PS : some modifications on the samples are not related to the patch I applied. Were the samples up-to-date ?
* Corrects issue #3410 when trying to create Arrays of primitive types
* Use c++11 nullptr keyword and various indentation issues resolved
* ran petstore on new mustaches
