[C++][Qt5] Rewrite of Qt5 Code Generator

[tbe]

The Code generated by the qt5cpp generator is incomplete and does not really match the other client library structures ( like cpprest, perl, ... ).

I would create a complete new set of templates and adapt the generator code, to make sure, the Qt5CPP generator behaves like cpprest and is more feature complete.

Is there any active maintainer for Qt5 or such a project already in progress?

As there is also #3203 outstanding for Qt5 i would also take care of this.

[wing328]

@tbe thanks for offering help to improve the Qt5 C++ generator.

I would suggest you to base the enhancements on 2.3.0 (for breaking changes)

I've copied the contributors of Qt5 C++ generator below:

@Ragnis @fehguy @sabras75 @philicious @fortiumant @marache

Ref: https://github.com/swagger-api/swagger-codegen/pulls?utf8=%E2%9C%93&q=is%3Apr%20label%3A%22Client%3A%20C%2B%2B%22%20is%3Aclosed%20QT5

[marache]

Hi @tbe, we have a current WIP version of a closer to Qt5 way of doing things, I'll try to propose this has a Pull Request linked to this issue. It might be a great way of starting a discussion ?

[tbe]

Hi @marache , that would be great.

@wing328, @marache i think, using QNetworkAuth for handling OAuth/OAuth2 as an optional feature would be a good idea. Any thoughts about this?

We could just ignore the authentication key handling for OAuth/OAuth2 based services and allow the user either to handle the complete authentication flow with the generated code, or just supply the authkey to the underlying QAbstractOAuth2 class.

[marache]

It's not a PR yet as we did a new plugin, I pushed the latest version here : https://github.com/cstb/qt5-swagger-codegen

[wing328]

i think, using QNetworkAuth for handling OAuth/OAuth2 as an optional feature would be a good idea. Any thoughts about this?

Never use it myself. After reading the doc, I think we should give it a try 👍

[tbe]

I'll take the code from @marache as base, integrate the authentication handling, better namespace support ( like cpprest ) and apply the google code style where needed and possible.

[wing328]

@tbe just want to let you know that we target to release 2.3.0 by the end of July (postponed from end of May earlier)

If the refactor is ready, we'll include it in the 2.3.0 release. (this is not to say we must include the refactored Qt5 c++ client in the 2.3.0 release so please take your time with the factoring and let us know if there's anything we can help)

[wing328]

FYI. We've started publishing a SNAPSHOT version of current master (2.2.3) and 2.3.0 (a branch with breaking changes). Please refer to https://github.com/swagger-api/swagger-codegen#compatibility for more information.

[antis81]

Once again I took the rocky road and went through closing all those nasty mem-leaks and thread leaks manually in the generated code base. Thanks to valgrind this is a doable task in hours instead of days… However there's not much needed to fix the default templates for good and make them "just work".

[antis81]

Do we have a 2.2.x branch where I can request the fix to be merged? It's not completely finished/tested, code is here -> antis81/patch-1

[wing328]

@antis81 please file against 2.3.0 branch, which will be released by the end of this month.

[antis81]

@wing328 I couldn't find a 2.3.x branch to file against. Do you mean master?

[wing328]

@antis81 right, I'm referring to the master (which will be released as 2.3.0). Sorry for any confusion.

[antis81]

Alright, rebased the changes onto master and pushed. Not sure everything works as I cannot actually test it. Could someone review or even test it further? (will copy the technical committee ^^) Am just about to open the PR and we see further… :)