These are chat archives for PySide/pyside2

28th
Apr 2017
From IRC (bridge bot)
@FromIRC
Apr 28 2017 09:02
<marco_piccolino> fkleint: any plans to officially integrate and support PyOtherSide as well at some point?
From IRC (bridge bot)
@FromIRC
Apr 28 2017 09:17
<fkleint> marco_piccolino: nope, what is this?
<fkleint> ;⁠-)
<marco_piccolino> fkleint: async Python bindings for QML: https://github.com/thp/pyotherside
<marco_piccolino> "async" and "Python" ... lovely words aren't they?
<fkleint> in the mix with QML
<marco_piccolino> indeed
<fkleint> alcroito, ctismer, ablasche: ^^^^^^^^
<alcroito> fkleint: yeah i saw it some time ago
<ctismer> me too
<marco_piccolino> looking forward to a Python sandwich stuffed with QML proteins
From IRC (bridge bot)
@FromIRC
Apr 28 2017 09:37
<fkleint> alcroito,ctismer: I think I found what is causing the MSVC test compile fail: https://codereview.qt-project.org/#/c/193039/
<fkleint> just need to test..
<alcroito> fkleint: how does that explain the flakyness of the building?
<fkleint> alcroito: I hope
<alcroito> fkleint: i'm asking "how"? :⁠D
<fkleint> that is, no, not really, but it is the compile error
<fkleint> I drilled it down to a qmake build
<fkleint> and found the wrapper lib causes 20000 unused variable warnings
<fkleint> which slowed Creator down to a cral;
<fkleint> which slowed Creator down to a crawl
<alcroito> let me rephrase the question. If we assume that the patch will fix the failing windows build, why did it sometimes fail, and sometimes not? How is defining noexcept to an empty string, sometimes works, and sometimes doesn't.
From IRC (bridge bot)
@FromIRC
Apr 28 2017 09:43
<fkleint> alcroito: Well, in the PySide world, there some things in between heaven and earth...
<alcroito> ok : ))
<fkleint> lemme just run the tests...
From IRC (bridge bot)
@FromIRC
Apr 28 2017 09:49
<fkleint> alcroito: Works, submitted. Can you try to re-run the CI?
<alcroito> fkleint: http://coin/coin/integration/pyside/pyside-setup/tasks/web_pyside_pyside-setup_1493373055276 dunno why the public facing coin doesn't work
<ctismer> Ah, forgot! This one has conflicts : https://codereview.qt-project.org/#/c/192553/
From IRC (bridge bot)
@FromIRC
Apr 28 2017 12:49
<ctismer> alcroito done!
<alcroito> @ctismer: yep, i'm running the tests
From IRC (bridge bot)
@FromIRC
Apr 28 2017 12:58
<fkleint> hrmpf: history out of range. Try 'testrunner.py list'
<fkleint> as I said there some things in between heaven and earth...
<alcroito> yep
<ctismer> fkleint: hä?
<alcroito> my suspicion is that coin somehow archives the submodule builds together with the build_history folders, and when it tries to parse those histories from a different build, it fails
<alcroito> but all these are assumptions without looking at the code yet : )
<fkleint> @ctismer: ^^^ COIN test log
<ctismer> are there multiple concurrent builds? That does not work.
From IRC (bridge bot)
@FromIRC
Apr 28 2017 13:03
<alcroito> well they are concurrent, but on different VMs
From IRC (bridge bot)
@FromIRC
Apr 28 2017 13:10
<ctismer> I think CI simply stores everything AND RESTORES everything. That is, the current status of the history
<ctismer> is sometimes reloaded with the project, but the current status is thrown away.
<ctismer> We must make sure that an additional store process is done after the build.
<ctismer> or maybe the history is simply ignored?