00:14 <isd> Ack.

00:15 <isd> The sched stuff I think I've seen before; though I thought we'd solved it (I can't remember how though, so I'm probably just wrong).

00:15 <isd> What it looks like is the install step for the test app is failing, because it expects a 'confirm' button to show up that never does, because the app is already installed

00:16 <isd> Probably we should tweak that helper so that it doesn't "demand" the confirm button.

00:17 <kentonv> yeah I noticed it only failed on the second test, probably because the app is already installed

00:17 <isd> I'm not surprised about the meteor stuff... I kinda had a feeling of "I can't reproduce the bug anymore but I'm not convinced it's fixed" at the time

00:17 <kentonv> but I couldn't actually find where the uploadTestApp() helper is defined... it doesn't show up in `git grep uploadTestApp`... O.o

00:18 <kentonv> well, I pushed a release despite the test failures since both failures look like problems with the test

00:19 <isd> Are we pulling anything in as sub-modules that might define it? libraries we wrote that are in separate repos?

00:19 <isd> That's bizzare.

00:19 <kentonv> nope

00:19 <isd> Yeah, I see three call sites and no definition sites. wtf?

00:20 <kentonv> oh there's a file called tests/commands/uploadTestApp.js

00:20 <kentonv> I guess it must come from the file name?

00:20 <kentonv> how strange

00:20 <isd> I hate javascript somtimes

00:20 <isd> But yeah, this looks like it.

00:21 <isd> I'll get a fix out soon.

00:26 <kentonv> no rush

00:28 <isd> The go thing appears to be an epoll failure

00:29 <isd> How does epoll interact with fuse generally?

00:42 <kentonv> it... really shouldn't

00:42 <kentonv> there's no reason to epoll a disk file

00:42 <kentonv> and there's no reason a FUSE file should behave differently when epolled

00:54 <isd> It looks like they just have one giant epoll() managed by the scheduler. It's going through a bunch of code that's generic to FDs, so it seems likely they're doing it that way just to avoid having an extra code path.

00:54 <isd> https://github.com/golang/go/blob/go1.13.7/src/os/file_unix.go#L152

00:56 <isd> So it looks like they're assuming the kernel will still tell them "yes it's readable" in spite of the error, but instead it's giving us this error? But why only with the fuse fs?

01:29 <isd> Ok staring at it some more I don't think the intent is for it to hit the epolling codepaths in the case of disk files. But the only way that error gets returned is if at some point the poller sets a flag for that file so... more investigation I guess.

02:01 <crab> hi

02:03 <isd> Ahoy

02:50 <isd> My brain has been moving slowly today.

02:51 <isd> spent a good chunk of the afternoon staring at Go's runtime trying to figure this out: https://github.com/sandstorm-io/sandstorm/issues/3219

02:52 <isd> My mind is immediately jumping to the experimental 3d file manager thing that was featured in jurassic park

02:53 <isd> https://en.wikipedia.org/wiki/Fsn_(file_manager)

02:58 <isd> Is anyone doing anything with webvr?

03:01 <isd> I'm going to take a shower. later.

03:37 <isd> Yeah, it's basically the MS-paint of voxel editors.

03:38 <isd> My experience has been that unless there's something inherently 3d to the subject matter, 3d interfaces on a 2d screen tend to be kinda gimmicky.

03:39 <isd> I think they have the potential to be less so given a headset, and I've seen some neat demos.

03:46 <isd> xet7_: hear that?

03:52 <isd> tempted to try out this with "cardboard" headset: https://github.com/immersive-web/webxr-polyfill

03:53 <isd> I'm skeptical of how much I think can fruitfully be tacked on to the basic web kanban board design.

03:58 <isd> We'd probably need to do "full screen grains" before you could fruitfully mess with webxr apps in sandstorm using that polyfill though.

04:02 <isd> Actually, apps can already probably do what they need with the browser fullscreen apis; they don't need to pop out of the iframe I think.

04:04 <isd> So there you go! start using webxr on sandstorm now!

14:03 <ocdtrekkie> kentonv: We should probably get Adam's test Action merged if we're gonna be hacking on fixing tests!

14:03 <ocdtrekkie> It looks like the failures his Action had match the test failures you documented here.

14:04 <ocdtrekkie> Also, if you pushed a release yesterday, I don't think it made it to GitHub.

14:38 <kentonv> whoops sorry, pushed now

14:41 <ocdtrekkie> Yay, closed an issue. \o/

14:42 <kentonv> ocdtrekkie, are you sure? Note I updated to Meteor 1.8.2 but not 1.9...

14:43 <ocdtrekkie> kentonv: Yup. I looked for the github-oauth package update.

14:43 <kentonv> yay

14:44 <ocdtrekkie> Warning every three days is an... irritating choice by GitHub for something deprecated five months from now.

14:44 <ocdtrekkie> Although I suppose it's an effective way to ensure everyone gets off dependencies of it. :P

14:45 <ocdtrekkie> vagrant-spk's release process will have the same problem, with github-release, but I am going to try to migrate to Hub, one of GitHub's own CLI tools.

16:40 <ocdtrekkie> Aaaaaand... building the Windows EXE does not work. I will see if I get time later today to troubleshoot it.

16:41 <ocdtrekkie> Something about the python.exe file being invalid.

16:42 <ocdtrekkie> I think I installed the wrong wine? The install command for wine defined in the Makefile was ambiguous and required I select my own.

19:17 <ocdtrekkie> So, it appears to be near impossible to install wine32 on Ubuntu 18.04.

19:18 <ocdtrekkie> Every advice, every third party repo I've tried, failure.

19:24 <ocdtrekkie> I'm considering spinning up a 32-bit Ubuntu VM on my Windows machine just to try to correctly install 32-bit WINE on Ubuntu so I can pack this Windows EXE.

19:32 <isd> Is there a reason not to use 64-bit wine?

19:46 <ocdtr_web> isd: Without wine32 it doesn’t support 32-bit EXE files.

19:46 <ocdtr_web> A large portion of Windows binaries, if they don’t need to be 64-bit, are only released 32-bit. Which is presumably why the script requires it.

19:47 <ocdtr_web> When I did a normal ‘apt install wine-stable’, it did not understand how to read python.exe

19:50 <isd> It seems like if it would be easier, you may as well just make it 64-bit. 32-bit only systems are pretty rare these days.

20:01 <ocdtr_web> But in the Windows world, a lot of tools like those used to pack this EXE may not even have 64 bit versions. I’ll have to check if that’s even an option.

20:02 <ocdtr_web> It’s unusual to Windows folks to be in an environment where 32-bit things don’t ‘just work’.

20:02 <ocdtr_web> I have some other stuff to do today but I’ll see if I can fight it a bit more tonight.

20:03 <isd> Whatever makes it easier.

22:40 <xet7> isd: When using some VR, in that VR 3D space just setup multiple browser windows in that unlimited amount of space. I have not tried VR yet.

22:41 <xet7> isd: It could be like big control room full of displays

22:42 <xet7> isd: I have not tried to make separate VR version of Wekan yet. It would be possible to do it with Godot. Depends if there is a real need to throw around cards in 3D space.

22:43 <xet7> isd: I also don't know does VR make using Wekan any easier, or do hands just get more tired when throwing cards around.

22:46 <xet7> isd: Compared to current Wekan 2D in browser, 3D would required more powerful computer to handle the environment, and VR is 2 x 3D - one image for each eye, it takes a lot of CPU power

22:48 <xet7> isd: There has been some feature requests for arbitraty placement of cards like a mindmap, and some other crazy features requests, that would be quite compilicated to do with webbrowser.

22:48 <xet7> isd: But with 3D engine like Godot it would be possible.

22:49 <xet7> isd: Godot can mix 2D and 3D graphics

23:53 <ocdtr_web> isd: Inno Setup is 32-bit only, I think.

23:53 <ocdtr_web> Which means, I either need to figure out how to fix the unholy horror that is WINE on 64-bit Ubuntu 18.04 or get a new Linux install.