#pyqt | Logs for 2019-01-03

Back
[00:11:40] -!- Belxjander has quit [Quit: AmigaOSv4.1.6+//PowerPC native]
[00:12:08] -!- Belxjander has joined #pyqt
[00:36:48] -!- Belxjander has quit [Quit: AmigaOSv4.1.6+//PowerPC native]
[00:37:29] -!- Belxjander has joined #pyqt
[01:26:36] -!- mintograde has quit [Ping timeout: 272 seconds]
[01:58:02] -!- kallesbar has joined #pyqt
[02:20:18] -!- JanC_ has joined #pyqt
[02:20:19] -!- JanC has quit [Read error: Connection reset by peer]
[02:21:59] JanC_ is now known as JanC
[04:12:37] -!- anqxyr has joined #pyqt
[05:56:04] -!- Belxjander has quit [Quit: AmigaOSv4.1.6+//PowerPC native]
[05:56:29] -!- Belxjander has joined #pyqt
[06:38:08] -!- BPL has joined #pyqt
[09:40:58] -!- thebigj has parted #pyqt
[10:05:22] -!- linuxson has joined #pyqt
[10:37:29] -!- linuxson has quit [Ping timeout: 246 seconds]
[11:23:28] -!- russw has joined #pyqt
[12:03:45] <russw> I'm trying to get the pyqt5qmlplugin.dll (from pyqt5-tools) working on Windows, but ran into an error. Has anyone tried this, or had success with it?
[12:03:53] <russw> The oh-so-useful error:
[12:03:54] <russw> https://bpaste.net
[12:04:56] <altendky> russw: i forget how to use it... i was going to add some helper commands to document and implement how to work with it but haven't yet
[12:05:49] <altendky> russw: i _think_ that with a little copying of files etc we (Siecje isn't here at the moment) made it do something 'useful'? but it's been a few months...
[12:06:04] <russw> I've used the linux version before with no problems... just struggling with windows
[12:06:17] <russw> I know Siecje well, and he never got it workign on Windows either
[12:06:23] <altendky> russw: and i'm making no promises it's built properly even.
[12:06:29] <altendky> russw: what qmlscene are you running?
[12:06:33] <russw> altendky: fair enough :)
[12:07:13] <russw> It occurs to me I should check your build on linux and see if that works
[12:07:33] <altendky> russw: naw, i've got a linux branch but... meh
[12:07:41] <altendky> i need to redo it
[12:07:56] <altendky> and it wasn't working. i hadn't gotten through the collection of applications iirc
[12:08:12] <russw> In the past I've built PyQt on linux to get the .so, but I've been beating my head on my desk trying to build PyQt5 on windows with no luck so far.
[12:08:30] <russw> altendky: why meh on the linux version?
[12:08:39] <russw> what platform are you normally on?
[12:08:49] <altendky> russw: because i started it and didn't actually get it to even an alpha state
[12:09:00] <altendky> russw: i work in linux. but i've only really done pyqt5-tools for windows
[12:09:07] <russw> ah
[12:09:24] <russw> ok, I won't bother checking yoru linux build then
[12:09:26] <altendky> russw: i did it so coworkers could use my custom widgets for dash layouts in designer. they mostly don't
[12:09:52] <russw> my goal is qmltestrunner, fwiw
[12:09:53] <altendky> russw: no idea why it's so hard... also, no idea why it's so important for phil to not provide an extra package with these things built
[12:10:24] <altendky> russw: but, phil is working on a setup.py so at least there will be a public example (sort of) of a 'build script'
[12:10:43] <altendky> as opposed to just docs and no full build process we can look at which is known to work
[12:10:45] <russw> I don't know why he doesn't include them in the main wheel. It seems to me like most people installing with pip will be devs who want a bit more.
[12:11:34] <altendky> russw: i'm fine with separation of deployment pieces vs. dev pieces. though honestly if you want to make deployments smaller, separate out the web stuff! that's gotta be a couple orders of magnitude larger than the designer and qml plugins
[12:12:06] <russw> what are you building with? Visual Studio, or MinGW?
[12:12:49] <altendky> russw: vs. though i honestly don't even understand how much the plugin needs to interact with the 'regular' pip installed pyqt5
[12:13:21] <russw> me either. It's kind of voodoo to me. Unfortunately.
[12:13:24] <russw> Drives me nuts that it is.
[12:14:56] <russw> altendky: which VS version are you using?
[12:15:18] <altendky> russw: should be the proper one for the python that is being built for. let me find the code
[12:15:52] <altendky> russw: https://github.com
[12:15:59] <altendky> i think i was cheating on one though...
[12:16:12] <altendky> yeah, python3.7 32-bit
[12:16:40] <altendky> get 14.0 rather than 14.14. i think 14.14-32 doesn't exist, iirc? and 14.0 is supposed to be compatible anyways
[12:17:33] <russw> I never even got a sip build to work on Windows... but wasn't matching versions. Maybe that was the issue.
[12:18:13] <russw> 3.7 32-bit is what I'm after... could that cheating be my problem?
[12:18:29] <altendky> russw: it'd be worth trying something else i'd say
[12:18:43] <altendky> russw: fyi https://wiki.python.org
[12:18:51] <altendky> doesn't have 3.7 though
[12:19:05] <altendky> you can open the repl and it should say near the top iirc
[12:19:18] <russw> altendky: this was my pip install:
[12:19:18] <russw> https://bpaste.net
[12:19:36] -!- thomasross_ has joined #pyqt
[12:19:36] -!- thomasross has quit [Killed (cherryh.freenode.net (Nickname regained by services))]
[12:19:36] thomasross_ is now known as thomasross
[12:19:52] <russw> my python build:
[12:19:53] <russw> Python 3.7.1 (default, Dec 10 2018, 22:09:34) [MSC v.1915 32 bit (Intel)] :: Anaconda, Inc. on win32
[12:20:27] -!- thomasross has quit [Max SendQ exceeded]
[12:20:40] <altendky> russw: https://en.wikipedia.org
[12:20:52] <altendky> so 14.15. hmm...
[12:20:55] -!- thomasross_ has joined #pyqt
[12:20:55] thomasross_ is now known as thomasross
[12:21:28] <russw> so weird that 3.7 isn't on that compiler list
[12:22:50] <altendky> russw: maybe bring it up in #python-dev or maybe #python. i thought i did at some point but i forget
[12:23:28] <russw> that list is presumably only for their stock binaries, anyway
[12:23:43] <russw> my anaconda build is for sure VS 14.15 as you say
[12:23:46] <altendky> russw: sure. you couldn't really have a list for much else.
[12:23:52] <altendky> russw: oh... i totally missed the anaconda
[12:23:53] <altendky> :[
[12:23:57] <altendky> russw: yeah, why anaconda?
[12:24:08] <russw> convenience of minconda
[12:24:15] <russw> *miniconda
[12:24:18] <altendky> russw: what is convenient about it?
[12:24:24] <russw> why the frown?
[12:24:39] <altendky> russw: 1) i missed it 2) it's different
[12:24:44] * altendky shrugs
[12:24:47] <russw> altendky: I like the multi-version handling and just generally like conda envs
[12:25:09] <altendky> russw: can you have conda envs use python.org pythons?
[12:25:13] <russw> for a while I was running a range of python versions and conda runs it well
[12:25:16] <russw> I even use it on linux
[12:25:28] <altendky> russw: i pyenv on linux
[12:25:31] <russw> altendky: no lcue about using python.org binaries
[12:25:38] <russw> *clue
[12:25:51] <altendky> russw: regardless, if you are digging on this i'd at least take the time to install the python.org python and see. should only take a couple minutes
[12:25:58] <russw> sure
[12:26:12] <russw> I might try pyenv as well. Been meaning to anyway.
[12:26:18] <altendky> russw: maybe do both 32/64
[12:26:30] <russw> pyenv can handle that... correct?
[12:27:32] <russw> nevermind... no good on Windows
[12:27:37] <altendky> russw: pyenv isn't so much a windows thing.
[12:27:59] <altendky> russw: pyenv builds python for you. on windows you mostly want to use the expected python.org stuff
[12:28:10] <altendky> or at least a compatible msvc build
[12:28:42] <russw> trying the official builds
[12:30:22] <altendky> russw: https://github.com hmm, not very useful :[ though i _think_ that the (linked) official docs mostly covered the steps. i just though they were super silly and annoying and so wanted to wrap them up with a cli to implement them
[12:30:34] <altendky> stuff like having to have a separate copy of the dll for each place you use it
[12:30:47] <altendky> and setting the import path etc
[12:31:39] <russw> I already have that issue open and was going to post success there when it worked, with an example of how to make it work on Windows.
[12:31:42] <altendky> russw: but i did make two entry points for qml stuff v
[12:31:44] <altendky> https://github.com
[12:31:51] <altendky> russw: :] thanks
[12:32:05] <altendky> https://github.com
[12:32:11] <altendky> https://github.com
[12:32:19] <russw> I was looking at the pyqt5* executables, and it isn't clear to me what they are for (aside form that .env thing you include)
[12:32:28] <russw> *from
[12:32:44] <russw> ah... thos elinks help
[12:32:46] <altendky> russw: the .env and being able to specify the paths as command line options
[12:33:05] <altendky> rather than having to figure out how to do env vars in whichever of the umpteen shells you might be in
[12:33:16] <altendky> (once supporting linux etc, or even just powershell vs cmd)
[12:33:40] <russw> it definitely sucks that you can't do `MY_ENV_VAR=123 go.exe` in Windows
[12:33:46] <russw> man I hate working in Windows
[12:35:47] <russw> I don't speak click... what are your default QML2_IMPORT_PATH values? Is "." in there?
[12:36:41] <altendky> russw: it just adds whatever you pass to the existing var (or creates the var if needed) QML2_IMPORT_PATH
[12:36:42] <altendky> ett
[12:36:43] <altendky> err
[12:36:47] <altendky> https://github.com
[12:37:07] <altendky> (copy/pasted between the two :[ )
[12:37:17] <russw> I just can't tell what populates qml2_import_paths
[12:37:46] <altendky> russw: a list created from each time you pass --qml2-import-path
[12:37:50] <russw> ah... seems to be -p
[12:37:56] <altendky> or -p
[12:38:05] <altendky> https://github.com
[12:38:20] <altendky> multiple=True so you can pass multiple. either --qml2-import-path or -p will feed it
[12:38:36] <russw> that *is* convenient.... I was planning a batch file
[12:38:36] <altendky> it will verify that it is a directory (not file) and that it exists
[12:38:56] <russw> anyway... step 1 is getting the hello_world of qml plugins working
[12:38:57] <altendky> russw: if you come up with things that would be useful to add, i'm game.
[12:38:59] <altendky> also, the .env
[12:39:08] <russw> I'll try with stock python and report back
[12:39:15] <altendky> russw: thanks for digging on this
[12:39:22] <russw> np :)
[12:39:49] <altendky> russw: maybe also make sure the designer plugin works
[12:39:52] <russw> if this works it is *much* more convenient than having to rebuild pyqt all the time!
[12:40:12] <russw> I've never even used the designer plugin, so I'm not sure how useful I'll be there
[12:40:36] <altendky> `yourenv\scripts\pyqt5designer --example-widget-path` and make sure that designer launches and you can see a (pointless) extra group and widget at the bottom of the widget list on the left
[12:40:58] <russw> I'll try it
[12:41:24] <russw> no need for anything but that? I'll just do it now as well... hang on.
[12:41:25] <altendky> both so i get more feedback about it working or not, but also so you have a reference as to whether the pyqt build i'm doing is working at all in your particular env
[12:42:04] <altendky> russw: yeah, i specifically included the sample widget and the cli option to make it easy for people to see if it's their fault or mine (or at least a first pass at figuring that out)
[12:42:17] <russw> My current build (where the qml plugin did not work) seems to have worked... proof in 10 seconds
[12:43:20] <russw> launch output: https://bpaste.net
[12:43:50] <russw> success: https://i.snag.gy
[12:45:08] <russw> so... your build/compiler matching works for the designer, but not the qml plugin
[12:45:35] <altendky> russw: thanks
[12:46:19] <russw> altendky: anything else needed to test that designer plugin, or is the appearance of the widget enough?
[12:48:38] <altendky> russw: that's it
[12:48:44] <russw> ok
[13:02:22] <russw> altendky: where I'm at so far:
[13:02:23] <russw> https://bpaste.net
[13:02:41] <russw> I tried your pyqt5qmlscene launcher to start. I'll try more.
[13:05:25] <altendky> russw: it didn't seem to get your import path
[13:05:34] <altendky> Try the cli parameter?
[13:05:51] <russw> I'm switching out of powershell and trying other stuff... hang on
[13:06:04] <russw> (and what is the cli parameter?)
[13:08:18] <altendky> russw: -p or --qml2-import-path
[13:08:37] <altendky> russw: you saw the print of `QML2_IMPORT_PATH: ;` ?
[13:08:43] <russw> oh right. thought you meant -xli
[13:08:46] <russw> er... -cli
[13:09:45] <russw> with -p:
[13:09:45] <russw> https://bpaste.net
[13:12:35] <altendky> russw: i wish i had logs from when siecje and i were talking about this...
[13:12:45] <altendky> i'm not certain if we got further than this or not
[13:13:39] <altendky> looks like september 11
[13:13:50] <russw> I know siecje... I'm certain you didn't. :)
[13:14:20] <russw> I'm still poking around. This also may be related to when we dug into this for linux (this thread):
[13:14:35] <russw> https://riverbankcomputing.com
[13:18:18] <altendky> huh, irccloud keeps logs a lot longer than i thought (at least for low traffic channels)
[13:18:43] <altendky> russw: virtualenv or venv?
[13:18:57] <altendky> virtualenv has had various issues with similar symptoms
[13:19:08] <russw> neither right now. just using the py launcher
[13:19:19] <russw> (a windows thing)
[13:19:29] <altendky> i'm familiar :]
[13:19:40] <altendky> dev in linux but deploy to windows
[13:19:44] <altendky> mostly
[13:20:07] <russw> I wish I knew who was throwing "The specified module could not be found." ... is that Qt, PyQt, or Windows?
[13:20:41] <russw> seems to be windoes
[13:20:44] <russw> windows
[13:22:13] <altendky> russw: i've dug up various tools while trying to debug the designer plugin but nothing is consistently useful
[13:22:14] <russw> It is comical how many links for windows errors say stuff like "update your drivers, run antimalware scans, reboot, reinstall windows, etc"
[13:22:50] <russw> I'm looking for strace-like tools, but expect that that won't actually help much
[13:23:06] <altendky> https://docs.microsoft.com
[13:23:17] <altendky> though i think i mostly used that to see the qdebug output
[13:23:21] <russw> of course it is sysinternals
[13:23:28] <altendky> http://www.dependencywalker.com
[13:23:39] <altendky> but it's old and i think that's starting to actually cause issues
[13:24:01] <russw> I used it recently and it was still quite helpful
[13:24:10] <altendky> https://github.com
[13:24:18] <altendky> russw: sometimes for some things, yeah
[13:24:21] <russw> actually fro pyinstaller stuff for pyqt and qt/qml stuff... useful for identifying needed dlls still
[13:27:52] <altendky> russw: so this had us a bit worried https://github.com
[13:28:00] <altendky> it didn't seem like the designer plugin did that
[13:29:16] <altendky> one of my wrapper ideas: 14:14 <altendky> i suppose for each directory in QML2_IMPORT_PATH i search each subdirectory for a qmldir file and check that file for pyqt5qmlplugin and if it's there i copy the dll to that directory. there would be a copy cli flag that'd default to on but could be turned off. or maybe the other way
[13:30:43] <russw> dependencywalker is losing its mind over that pyqt5qmlplugin.dll
[13:30:53] <altendky> ms-api- stuff?
[13:30:58] <altendky> or whatever it is
[13:31:02] <russw> still waiting
[13:32:07] <russw> seems like it could have been recursion issues in the kernel32 walking
[13:32:15] <russw> anyway, here is the end result (collapsing deep stuff)
[13:32:16] <russw> https://i.snag.gy
[13:32:49] <altendky> russw: it doesn't understand the api-ms indirection so i think that's not meaningful either way?
[13:33:02] <altendky> and i don't even remember how to read the rest
[13:33:16] <russw> agreed. those api-* dlls are funky
[13:33:30] <russw> so dependencies seem ok I think
[13:33:57] <altendky> russw: do we have evidence of this qml stuff working 'recently'?
[13:34:06] <altendky> russw: you said you do or do not have it working in linux?
[13:35:18] <russw> I used to have it working on linux. I've not verified it recently, although siecje and I tried to revive it just before the holidays.
[13:35:45] <altendky> russw: so we had some concern about whether it needed updated after the sip changes
[13:35:46] <russw> I need windows working, so I went straight to windows. If this dead-ends, I'll confirm linux functionality first.
[13:36:31] <altendky> this is the issue with incomplete public binary release and no public test results... we literally just don't know if it works _anywhere_. *sigh*
[13:36:44] <altendky> it could be utterly and entirely broken
[13:36:58] <russw> it being damn near impossible to compile from scratch on windows isn't helping either
[13:37:35] <russw> nor is the fact that the examples *are* (or were) completely broken in the first place. Took some serious hoop jumping for it to work on linux in the past.
[13:37:40] <altendky> i brought this up in some maillist email. phil's response didn't make it clear he understood the value of these things being public
[13:37:56] <russw> he also doesn't like qml
[13:38:13] <altendky> i mean just a public build script
[13:38:26] <altendky> and a build of it run publicly. or at least the log of that build being public
[13:38:27] <altendky> let alone test
[13:38:28] <altendky> s
[13:38:41] <russw> the fact that siecje and I are the only ones who seem to be using PyQt with QML (and we work together) is also a bit disturbing, but that is another issue... the sphere is small
[13:39:32] <russw> have you tinkered with pyside2 (Qt for python) yet? I'm hopeful for it, but... 'tis early.
[13:39:39] <altendky> maybe just ask the maillist and phil if anyone has run it recently at all
[13:39:59] <russw> siecje and I were literally the only people who cared in the past
[13:40:20] <altendky> russw: i asked them if they would support pypy. they responded it was too early to decide that... as if deciding if you use the cpython c-api isn't the _very_ first thing you decide. *sigh*
[13:40:27] <russw> but I will try that, after I poke around for a bit
[13:40:57] <russw> altendky: you mean for pyside2?
[13:41:02] <altendky> yes
[13:41:07] <altendky> russw: i haven't messed with it really. i'm curious as well. meejah is using it with twisted stuff and qt5reactor.
[13:41:47] <russw> I veered hard away from twisted, but use a standalone deferred taht I wrote a custom "QtDeferred" wrapper around. Works well to date.
[13:44:28] <altendky> being able to use @inlineCallbacks or for the past couple years async/await is nice
[13:45:05] <altendky> signals and slots fulfill one async need. they do not make it nice to code a sequential activity with several async 'disconnects' in the middle
[13:45:11] <altendky> especially not when you want to loop
[14:01:41] <russw> debugging "asynchronous" code is hellish, even with @inlinecallbacks (which rocks)
[14:02:26] <altendky> russw: not that it changes debuggability, but you know you can async/await with twisted?
[14:02:35] <russw> I did not
[14:02:40] <altendky> a couple years now
[14:02:55] -!- y0u has quit [Quit: Connection closed for inactivity]
[14:03:11] <russw> full-time 3.x is only relatively recent for me
[14:03:48] <russw> I've been considering rewriting code to support async/await, but am not sure it will add anything useful, other than using the new shiny thing
[14:04:29] <altendky> russw: probably mostly not. technically it relieves yield back to its other duties so you can have an async generator
[14:04:36] <altendky> but yeah, that's likely not the most common thing
[14:04:51] <russw> If breadcrumbs existed to support the "How the heck did I get here?" question that happens all the time with async coding, that would be *awesome*... frigging event loops.
[14:05:07] <russw> Right now I need extensive logging to support that. It works well, but is verbose.
[14:05:26] <altendky> russw: i tried eliot but ran into some issue in my case
[14:06:03] * russw looks up eliot
[14:06:26] <altendky> russw: but i would think that inline callbacks and async/await might help with having a 'call stack' as opposed to old-school .addCallback
[14:06:56] <altendky> https://eliot.readthedocs.io
[14:07:26] <russw> inlinecallbacks helps a huge amount for readability and debugging, but you still have painful async problems
[14:07:36] <russw> yeah, eliot looks neat... looking into it more
[14:08:22] <altendky> it's really silly that when i view an issue list on github i can't see what issues i've starred etc
[14:08:35] <altendky> or subscribed to rather
[14:09:01] <altendky> have to revert to google https://github.com
[14:09:40] <altendky> who knows, maybe async/await would help there
[14:10:38] <altendky> i'm vaguely planning on taking another stab at using eliot in my gui. i want an activity log which can be filtered for 'error's etc instead of just popping up modal dialogs when things happen
[14:13:32] <russw> eliot definitely looks cool. I've done various similar hacks in the past for nested logging, but that looks great.
[14:13:46] <russw> except for the "UNNAMED TASK" stuff with generators. ugh.
[14:16:23] -!- kallesbar has quit [Quit: Konversation terminated!]
[14:43:28] <russw> Am I supposed to be able to `import sip` in python after I compile sip with the "new" method: `python configure.py --sip-module PyQt5.sip --no-tools; make; make install`?
[14:43:44] <altendky> russw: no. `import PyQt5.sip` afaik
[14:43:57] <russw> ah...
[14:44:00] <altendky> the point is that it is build separately for each package that needs it
[14:44:08] <altendky> well, built, distributed, imported, etc
[14:44:14] <russw> I don't quite get the separate build thing here:
[14:44:14] <russw> http://pyqt.sourceforge.net
[14:44:49] <russw> import PyQt5.sip worked... thx
[14:48:24] <russw> hrm... sip.h doesn't seem to get put in the right place anymore:
[14:48:25] <russw> https://bpaste.net
[14:49:45] <altendky> Try again using
[14:49:45] <altendky> the --verbose flag to see more detail about the problem
[14:49:47] <altendky> :]
[14:50:33] <russw> I did. And I've just determined thereason that sip.h is missing. PyQt5 install docs say to use --no-tools, but...
[14:50:36] <russw> http://pyqt.sourceforge.net
[14:50:53] <russw> doing that specifically does not install the sip.h file needed for pyqt5 installation?!?!
[14:54:29] <russw> rebuilding sip without --no-tools seems to fix it
[14:54:38] <altendky> russw: yes, I find it funny that it told you to use verbose when you had used verbose...
[14:55:06] <russw> yep. been an "entertaining" build experience so far.
[14:56:19] <russw> care to place bets on whether this full linux rebuild will give me a functional libpyqt5qmlplugin.so? :)
[14:56:33] <AbleBacon> is there a simple way to preserve the text that was in a text box (or the "checked" or "unchcked" state of a checkbox) when a program is closed and then re-opened? i see some examples on google, but they look really complicated and hacky
[14:56:34] <russw> Vegas odds are 50/50
[14:57:18] <russw> AbleBacon: can you link an example? I'm curious.
[14:57:34] <AbleBacon> https://stackoverflow.com
[14:58:00] <russw> Any such "sticky" value will require coding something on your own. I haven't seen anything canned that will do it for you.
[14:58:01] <Avaris> AbleBacon: essentially, save somewhere. qsettings might be useful
[14:58:37] <AbleBacon> well, i know that they'll need to be saved in a location, but is there a built-in way to do it?
[14:59:00] <Avaris> yeah, qsettings ^ :S
[14:59:05] <AbleBacon> i'd prefer not to go behind the back of every widget and scape all of their properties and then try to serialize them, and vice versa
[14:59:22] <Avaris> oh like that, no
[14:59:23] <russw> AbleBacon: you can look at `appdirs` for python, or (since your using PyQt) you could use QSettings or QStandardPaths
[14:59:45] <AbleBacon> ok thanks i'll look into these
[14:59:46] <Avaris> you won't need to scrape *all* the properties though right?
[15:00:19] <AbleBacon> i mean, if you weren't scraping all of the properites, then you'd have to specify which ones to scrape which would make things even more difficult
[15:01:50] <Avaris> would it? i doubt. qt objects can have quite some stuff most of which i assume you don't care or need to save
[15:20:50] <russw> altendky: the latest released PyQt5 works fine with Qt 5.11.3 and its libpyqt5qmlplugin.so:
[15:20:52] <russw> https://bpaste.net
[15:21:12] <russw> (I have not looked into that sip attr error)
[15:24:27] <russw> altendky: this means that phil's current PyQt code does fundamentally work... at least on linux. Unless there is some Windows-specific difference that screws it.
[15:28:19] <russw> One point of interest is that if I don't hard set the LD_LIBRARY_PATH like I did, I get a "cannot load library" error, which is at least similar to the "specified module cannot be found" error that Windows gives
[15:28:28] <russw> see here:
[15:28:28] <russw> libpyqt5qmlplugin.so
[15:28:31] <russw> oops
[15:28:48] <russw> https://bpaste.net
[15:29:28] <russw> so maybe we just need the windows equivalent of setting LD_LIBRARY_PATH?
[15:39:36] <altendky> russw: i think it may just be the regular PATH?
[15:40:20] <altendky> AbleBacon: in general it's good to have your program implemented without a gui and have a gui be an interface to it. as is it seems your gui objects actually store your data.
[15:41:18] <altendky> AbleBacon: a qcheckbox for example has 72 qt properties http://doc.qt.io but presumably you only care about one of them
[15:41:34] <altendky> well, only care about one as far as persisting state across runs
[15:41:44] <altendky> the rest are more likely just gui setup stuff that's constant
[15:45:46] <altendky> russw: random newsgroup post but... https://groups.google.com!topic/comp.lang.tcl/J48G1yhvFrc
[15:45:55] <altendky> says to use PATH
[16:06:17] -!- Dave_Elec has joined #pyqt
[16:41:23] -!- nsizemor has quit [Quit: nsizemor]
[16:44:18] -!- Dave_Elec has quit [Ping timeout: 246 seconds]
[16:46:12] -!- Dave_Elec has joined #pyqt
[16:46:14] <russw> altendky: progress with PATH:
[16:46:15] <russw> https://bpaste.net
[16:46:59] <altendky> russw: isn't it great when errors are progress?
[16:47:08] <russw> :)
[16:47:16] <russw> but they are *new* errors! :)
[16:47:16] <altendky> a character changed... progress! :]
[16:47:26] <altendky> anyways, QtQuick...
[16:48:00] <altendky> do i need to be adding some paths under site-packages or whatever to the qml path?
[16:49:46] <russw> not sure yet
[16:51:20] <russw> making sure that the python dll is in the path does seem necessary, anyway.
[16:51:22] <russw> maybe
[16:51:33] <russw> easy to get with os.dirname(sys.executable), anyway
[16:52:13] <russw> or rather, os.path.dirname(sys.executable)
[16:59:11] -!- Dave_Elec has quit [Read error: Connection reset by peer]
[17:01:17] <russw> altendky: how can I tell which Qt dir your package selected as the correct one?
[17:01:37] <russw> Trying to determine which dir to include is educational:
[17:01:53] <russw> https://i.snag.gy
[17:03:34] <altendky> russw: it won't pick a qt dir. it would add the pyqt dirs in the env to the PYTHONPATH which it prints
[17:03:44] <altendky> or, well, the site-packages based on sys.path anyways
[17:03:55] <altendky> not sure if the plugin does anything more
[17:04:13] <russw> right... I realized it was a dumb question after asking :(
[17:04:28] <altendky> so either the site-package\pyqt5_tools or PyQt5\Qt\bin
[17:04:34] <russw> I just added the dir with all the Qt DLLs using the -p cli switch... but nope
[17:04:39] <altendky> theoretically they should be the same?
[17:04:50] <altendky> mm
[17:05:00] <altendky> which was that?
[17:05:15] <russw> https://bpaste.net
[17:05:22] <altendky> remember you wouldn't add the example Charts directory, you add the one above it. though that probably doesn't apply here i guess
[17:06:02] <russw> the way I have the local setup (in . and ./Charts) should be fine, since I'm mirroring my linux success
[17:07:16] <russw> this is tantalizingly close
[17:07:19] <russw> I think
[17:07:30] <altendky> russw: what about having it in PATH?
[17:07:40] <russw> worth a shot
[17:07:44] <altendky> doesn't make a lot of sense to me, but yeah, that
[17:07:45] -!- mintograde has joined #pyqt
[17:08:18] <russw> it's the QML import that is failing, so I figured QML2_IMPORT_PATH, but gimme a sec
[17:08:23] <altendky> i agree
[17:08:36] <altendky> does qmlscene take any debug/verbose options?
[17:09:42] <russw> nope on adding it to PATH
[17:09:45] <russw> checkign qmlscene
[17:10:20] <altendky> if so, pyqt5qmlscene should pass through
[17:10:51] <altendky> https://github.com
[17:11:06] <altendky> russw: and i do pass --qmlscene-help through as --help to qmlscene
[17:11:20] <russw> went direct and there is a --verbose
[17:11:25] <russw> didn't pass through your wrapper
[17:11:31] <russw> do you pass unknown switches through?
[17:11:37] <altendky> i thought i did
[17:11:51] <russw> you might. It might just not have provided anything interesting.
[17:11:56] <altendky> https://github.com
[17:12:18] <altendky> https://github.com
[17:12:22] <russw> you do
[17:12:29] <russw> all I got was the version, though
[17:12:32] <altendky> mm
[17:13:15] <russw> verbose, then not:
[17:13:15] <russw> https://bpaste.net
[17:13:36] <altendky> enough we see it's doing something at least...
[17:13:41] <russw> mismatch between 5.11.2 and 5.11.3 has me scratching my head, though
[17:13:56] <altendky> pyqt 5.11.3 uses qt 5.11.2
[17:14:12] <altendky> https://pypi.org
[17:14:21] <altendky> all the files are 5.11.3-5.11.2 which is saying that
[17:15:11] <altendky> have you grepped for the error message in case it's in a literal in the plugin code or somesuch?
[17:16:36] <russw> not this one. one sec.
[17:16:49] <altendky> i'm quite pessimistic but yeah
[17:27:16] -!- Belxjander has quit [Quit: AmigaOSv4.1.6+//PowerPC native]
[17:27:42] -!- Belxjander has joined #pyqt
[17:29:12] <russw> the "is not installed" message for sure comes from Qt. Funnily enough, the Charts plugin now loads ok, but not QtQuick. Confirmed by reversing the order of the imports and Charts imports ok.
[17:36:14] -!- AbleBacon has quit [Quit: Leaving]
[17:37:08] -!- thomasross has quit [Remote host closed the connection]
[17:37:34] -!- thomasross has joined #pyqt
[17:44:52] -!- anqxyr has quit [Read error: Connection reset by peer]
[17:57:08] <russw> altendky: fwiw, it is definitely something with your wrapper. See below (where hw.qml is a simple hello_world rectangle):
[17:57:11] <russw> https://bpaste.net
[17:57:35] <russw> the raw qmlscene works (of course), and the wrapped one gives the same error as the qmlplugin attempts.
[17:58:03] <russw> I'm hopeful that resolving this, and combining with the qmlplugin stuff from earlier, will give me the answer by end of day.
[17:59:26] <altendky> russw: i don't quite follow. it works with no python. it doesn't work with the pyqt qml plugin and my wrapper. so it's something to do with one of those two things, yes?
[17:59:43] <altendky> (i mean sure, i built the plugin as well so you can still point fingers at me regardless but... :] )
[17:59:46] <russw> one more bit of info: the qmlscene.exe *beside* your wrapper also does not work
[18:00:17] <altendky> russw: is there a QtQuick.dll (or whatever) next to my qmlscene.exe?
[18:00:34] <altendky> or in the same relative path as to the msvc2015\bin\qmlscene.exe
[18:01:04] <russw> all three here (only the first one works):
[18:01:04] <russw> https://bpaste.net
[18:01:32] <russw> checking your auwestions... hang on
[18:01:38] <russw> ?? *questions
[18:01:44] <altendky> russw: oh, add to the mix the actual qmlscene.exe from qt under site-packages\pyqt5_tools
[18:01:52] <altendky> auwesome questions
[18:01:59] <russw> :)
[18:02:13] <altendky> russw: the qmlscene.exe you show in the middle there under scripts is an entry_opints
[18:03:23] <russw> no adjacent dlls in the scripts dir
[18:03:43] <altendky> russw: right, that's a wrapper that runs the other qmlscene.exe
[18:03:52] <altendky> under site-package\pyqt5_tools
[18:04:16] <altendky> i generate functions and entry_points for each of the tools i include
[18:05:15] <russw> the direct qmlscene.exe doesn't work... where do you get it?
[18:05:25] <russw> Here is the ls of the scripts dir, and the failed direct execution:
[18:05:26] <russw> https://bpaste.net
[18:05:41] <altendky> russw: qt on appveyro
[18:06:04] <altendky> russw: this is 3.7-32?
[18:06:39] <russw> should be, anyway
[18:06:49] <altendky> russw: and i use the qt util to grab the files it depends on
[18:07:59] <altendky> https://ci.appveyor.com
[18:08:03] <russw> this is the python repl in the same dir I've been running your wrapper from:
[18:08:03] <russw> https://bpaste.net
[18:09:41] <altendky> russw: but i guess that site-packages qmlscene.exe is the closest thing to debug
[18:10:08] <altendky> russw: maybe cd to that directory and .\qmlscene.exe
[18:12:45] <altendky> then maybe a directory tree diff (meld and kdiff3 both do that) and maybe a dep walker/dependecies/whatever
[18:15:08] <russw> hrm: https://bpaste.net
[18:16:21] <altendky> russw: can you share the tree of C:\QT\5.11.3\msvc2015\bin ?
[18:17:00] <russw> the fact that the raw qmlscene.exe that you scoop doesn't work is great news to me. I think that getting this working may be the last thing.
[18:17:11] <russw> (he says with massive optimism)
[18:17:37] <russw> if thsi doesn't work, I don't expect the paired qmlplugin dll to work either
[18:17:43] <russw> weird that your designer works
[18:19:49] <russw> altendky: your scooped qtdiag.exe works, and may be informative:
[18:19:50] <russw> https://bpaste.net
[18:21:20] <russw> for example, it cant be good that the libraries path is `site-packages\pyqt5_tools\lib`, which does not exist
[18:21:36] <russw> I'm going to try skipping all of your wrapper stuff, but stealing the qmlplugin dll
[18:22:34] <altendky> russw: can you share the tree of C:\QT\5.11.3\msvc2015 ?
[18:23:07] <russw> altendky: it's huge
[18:24:36] <russw> altendky: dirs only:
[18:24:37] <russw> https://bpaste.net
[18:26:19] <russw> altendky: any idea how critical it is to match compiler version between 1) python, 2) QT, 3) sip, and 4) PyQT5?
[18:26:34] <altendky> russw: they need to be compatible
[18:26:52] <altendky> and pyqt is tied to a minor rev of python
[18:30:42] -!- BPL has quit [Quit: Leaving]
[18:33:40] <russw> so... what is "compatible"? My Qt libs that work are in \Qt\5.11.3\msvc2015\bin
[18:34:04] <russw> my python is built with vs2017 (specifically Visual Studio 2017 version 15.9)
[18:34:19] <altendky> 2015/2017 are supposed to be
[18:34:36] <russw> presumably my pyqt5 core release libs are compatible with the vs2017 python, assuming phil did that right
[18:34:42] <altendky> but is there not an msvc2017 for 5.11.3?
[18:34:50] <altendky> or, some other name for 2017
[18:34:50] <russw> not that I have
[18:34:56] <russw> I'll run the maintenance tool and check
[18:34:58] <altendky> it would be an install option
[18:50:59] <altendky> russw: https://www.appveyor.com
[18:51:33] <altendky> so there's msvc2017-64 but not -32 i think. which might be what i was saying earlier, i forget...
[19:13:03] -!- Belxjander has quit [Quit: AmigaOSv4.1.6+//PowerPC native]
[19:13:32] -!- Belxjander has joined #pyqt
[19:53:13] -!- thomasross has quit [Ping timeout: 258 seconds]
[19:54:04] -!- thomasross has joined #pyqt
[22:04:39] -!- thomasross has quit [Quit: Leaving]
[22:04:53] -!- thomasross has joined #pyqt
[22:08:42] -!- mintograde has quit [Ping timeout: 250 seconds]
[23:30:23] -!- mandeep has quit [Remote host closed the connection]