kentonv changed the topic of #sandstorm to: Welcome to #sandstorm: home of all things sandstorm.io. Say hi! | Have a question but no one is here? Try asking in the discussion group: https://groups.google.com/group/sandstorm-dev
nicoo has quit [Ping timeout: 240 seconds]
<xet7> Well, I hope Sandstorm uses Stylus 1.1.0, because building Stylus 2 takes 52 minutes https://github.com/wekan/wekan/commit/b3a7c4dc8867bf863fa5cca4e21b9c220026e76a
<xet7> after reverting to 1.1.0 Wekan builds in 3 minutes
<xet7> Hmm, I'll check
nicooo has joined #sandstorm
<isd> What is stylus? It doesn't look like we depend on it at all?
<xet7> Hmm, seems so
<xet7> stylus is converted to CSS
<xet7> .styl files
<xet7> Anyway, when building with Meteor, you can add METEOR_PROFILE=100 like this https://github.com/wekan/wekan/blob/master/releases/rebuild-release.sh#L12 to see what package takes most time to build
<xet7> And when building with newest Meteor 1.10.1, it's possible to exclude some archs like this: meteor --exclude-archs web.browser.legacy,web.cordova
<isd> Wait, are we building a mobile target for no reason?
<ocdtrekkie> I am trying to get .NET Core working in Sandstorm.
<ocdtrekkie> Four years later, I am now at the same spot the first person who tried to do it was at.
<ocdtrekkie> https://github.com/dotnet/runtime/issues/2534 if anyone wants to take potshot guesses at what part of the sandbox dotnet is mad about.
<TimMc> Progress! XD
<ocdtrekkie> I honestly did have some difficulty getting back to where he had gotten to! :P
<ocdtrekkie> I documented everything I played with mostly and made a test repo.
<ocdtrekkie> Hoping someone else can help me make some sense of it. Maybe in the dotnet repo.
<abliss1> taking potshots at obscure bugs where i have no domain knowledge is one of my favorite pastimes. but that error log doesn't give much to go on.
<ocdtrekkie> My curiosity is that, you know, I made a fake proc to do /proc/self/exe right?
<ocdtrekkie> I notice sandstorm-files.list has /proc/cpuinfo
<abliss1> maybe a TMPDIR problem? https://github.com/dotnet/runtime/issues/3168
<ocdtrekkie> I saw that too. But it's hard to like... narrow down which this error message could be.
<ocdtrekkie> Should I be setting the TMPDIR environment variable, or does it pick a reasonable default?
<ocdtrekkie> Is it because Sandstorm's /tmp is too small?
<abliss1> if you can follow the instructions on there to get an strace log, we'll likely be able to figure out which syscall failed and why
<abliss1> it probably picks /tmp by default. you could try setting TMPDIR to /var or something to see what happens.
<abliss1> "You could try disabling debugging with export COMPlus_EnableDiagnostics=0. There are some pipes that that are created in /tmp that this disables." -- sounds enticing
<ocdtrekkie> Gonna try the TMPDIR first, let me see where I get there.
<isd> Re: the build perf, I did the profiling thing that xet7 suggested, and it looks like the lion's share of the time goes into minifying javascript (and to a lesser extent css). Can we just turn that off for development?
<ocdtrekkie> TMPDIR got me nothing
<isd> I think strace depends on syscalls that are blocked by sandstorm :/
<ocdtrekkie> I don't feel like the COMPlus line helped either.
<abliss1> what's mounted at /tmp in the sandbox? any way to map a real tmpfs there?
<isd> it's a realy tmpfs, but it's capped at 16MiB
<abliss1> is there an easy way to try making it bigger? maybe dotnet's trying to put something large there?
<isd> No, it's hard-coded
<xet7> Wow. I just read from my Amiga 1200 some of the source code I did with Blitz Basic, coding a music note clicking game, from 1994.
<xet7> I did not remember I had coded that
<ocdtrekkie> abliss1: Setting TMPDIR=/var/tmp didn't help.
<ocdtrekkie> I really just need a less vague error message. But as this is Microsoft here I'm working with, an obscure error code with no explanation is par for the course.
<ocdtrekkie> I remember when Windows 10 came out, and upgrading failed, and you got the error message "Something happened"
<ocdtrekkie> No error code for that.
<isd> Yeah, that's kinda what drove me to Linux way back when; at least when something was broken I had a hope of figuring out why. I kinda moved away from the 'easy to use' distros for the same reason.
_whitelogger has joined #sandstorm
larjona has quit [Ping timeout: 264 seconds]
dckc has quit [Ping timeout: 246 seconds]
dckc has joined #sandstorm
frigginglorious has joined #sandstorm
frigginglorious1 has joined #sandstorm
frigginglorious has quit [Ping timeout: 240 seconds]
frigginglorious1 is now known as frigginglorious
nicooo has quit [Remote host closed the connection]
nicooo has joined #sandstorm
<isd> Which tests are (supposedly) still broken? I'd like to fix those so we can set CI to actually fail if we break more stuff.
TC01 has quit [Ping timeout: 250 seconds]
TC01 has joined #sandstorm
TC01_ has joined #sandstorm
TC01 has quit [Ping timeout: 265 seconds]
TC01_ has quit [Client Quit]
TC01 has joined #sandstorm
ocdtr_web has joined #sandstorm
<ocdtr_web> isd: Should be an issue tagged sandstorm-dev for each one.
ocdtr_web has quit [Remote host closed the connection]
<isd> Can we organize the CI breakage into a milestone? the sandstorm-dev label is useful but there's other stuff in there too, so it still takes some filtering.
<ocdtrekkie> I'm confused why Kenton's "Fix CI" commits definitely still leave CI broken.
<ocdtrekkie> I had to go REALLY far back to find a working build.
<ocdtrekkie> ✖ tests/account-settings ✖ tests/account ✖ tests/appHooks ✖ apps/web-publishing are the four currently persistent problem parties.
<isd> kentonv mentioned in the PR for apps/web-publishing (which isn't merged yet) that it's been broken since he implemented the gateway.
<ocdtrekkie> Yeah, I think I pasted his comment about it in the issue tagged in the milestone.
<ocdtrekkie> I also opened an issue for the error CI is currently falling flat on.,
<ocdtrekkie> #3249 looks like it was meant to help address some of these, but it was also broken.