shellac has quit [Quit: Computer has gone to sleep.]
<headius[m]> enebo: you already defer instruction decoding
<headius[m]> took me until now to find it but InterpreterContext gets created with a Callable that holds the original stream open
<enebo[m]> oh? maybe it was based on idea that most methods are lazy and you do not want to load the world
<headius[m]> see InterpreterContext.getEngine
<enebo[m]> I have wanted to write an IR assembler so that testing could be done decoupled from everything else (wrt your idea above)
<headius[m]> well this day wasn't a waste at least...I fixed some persistence bugs
<headius[m]> I've started trying to get it to test in spec:compiler but having some minor issues I'll pick up tomorrow
<enebo[m]> yes. AOT of anon kwrest would have crashed and you said a couple of others
<headius[m]> rails c is able to load from persistence without errors
ur5us has joined #jruby
<enebo[m]> I am going to debug my encode/decode of argumentdescriptors but may be done after that depending on how quickly I solve it
<headius[m]> given this I guess I can proceed to merging bytecode AOT with persistence AOT
<enebo[m]> yeah
<enebo[m]> and this step was interesting from a talk perspective as well
<enebo[m]> I was thinking about how maybe besides explaining why we care and why this important we could do some history on various attempts
<enebo[m]> headius: e31a5318861ab282e041857d9b190c629e04e94a
<enebo[m]> Do you feel old yet?
<headius[m]> I've push my changes... here's what I have for testing that NPEs for some reason: https://gist.github.com/headius/00242e561beb79785fb5bcb48ee21647
<headius[m]> wow 2016
<headius[m]> at least I know I have the same good ideas over again
<enebo[m]> After seeing the callback this sort of paged back in for me
<headius[m]> I guess I'm disappointed it doesn't show more gain but at least it is actually as fast or faster for everything I tested
<headius[m]> assuming it's working
<enebo[m]> Of course once you merge AOT + serialize this will never happen
<headius[m]> right
<headius[m]> and presumably some combination of JVM features will make the bytecode part cheap
<enebo[m]> yep
<headius[m]> so we're closer than I thought perhaps
<enebo[m]> and if my stuff turns out we will only be loading static scope
<headius[m]> worst case right now is that it does actually stand up the IRScope but it will be small
<headius[m]> yeah
<headius[m]> so progress
<enebo[m]> yep. I mean we don't know how well it will all work but we know it will be faster
<enebo[m]> I don't see how it could not be
fidothe has joined #jruby
<headius[m]> I'm calling it a day and I'll be off tomorrow so ttfn
<headius[m]> if you get bored see if you can get that testing working
<headius[m]> in that form or some other
<headius[m]> we're going to be relying on persistence
<enebo[m]> yeah we sort of already do too
<headius[m]> well, maybe...dunno how many people actually are using AOT right now
<headius[m]> they wouldn't get far
<enebo[m]> that was my "sort of"
<enebo[m]> I don't know what you fixed but ** and * with no args is fairly rare
<headius[m]> SetCapturedVarInstr was broken too
<headius[m]> it encoded a string and decoded a symbol
<headius[m]> and IRFor would not decode properly
<headius[m]> I'm not sure when SetCapturedVar is used
<enebo[m]> buildMatch2
<headius[m]> ah
<headius[m]> for the special locals
<headius[m]> yah so any script with a for or a match2 with named groups would have failed
<headius[m]> along with the KW_REST thing
<enebo[m]> it does really underscoe the need for tests regardless of how common they are
<headius[m]> anyway it's on master so you can take a peek
<headius[m]> yeah these were just minor missed things
<headius[m]> but never come up unless we roundtrip test
<enebo[m]> I know people use AOT because it is usually how I find out these are broken :)
<headius[m]> for all I know my NPE trying to test could be another bug
<headius[m]> hard to say, rspec messes with stuff
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<travis-ci> jruby/jruby (jit_irscope_removal:ef55e6b by Thomas E. Enebo): The build was broken. https://travis-ci.org/jruby/jruby/builds/637632613 [170 min 30 sec]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:c48536c by Charles Oliver Nutter): The build was broken. https://travis-ci.org/jruby/jruby/builds/637713551 [178 min 37 sec]
travis-ci has left #jruby [#jruby]
fidothe has quit [Read error: Connection reset by peer]
Liothen has quit [Ping timeout: 260 seconds]
den_d has quit [Ping timeout: 248 seconds]
Iambchop has quit [Ping timeout: 272 seconds]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (jit_irscope_removal:8f9bc92 by Thomas E. Enebo): The build is still failing. https://travis-ci.org/jruby/jruby/builds/637721104 [164 min 21 sec]
travis-ci has left #jruby [#jruby]
<rwilliams[m]> enebo: Work and play
* rwilliams[m] uploaded an image: image.png (2844KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/CGevGeztkubFDqnIDUpaPRTc >
<enebo[m]> rwilliams: nice. If I spend a few minutes cleaning I can show my desktop :)
<rwilliams[m]> Though honestly between WSL2 and dual boot ubuntu 19.10 I spend less and less time on the macbook pro. I'm seriously considering a Windows laptop for my next portable.
travis-ci has joined #jruby
<travis-ci> jruby/jruby (jit_irscope_removal:14207c1 by Thomas E. Enebo): The build was fixed. https://travis-ci.org/jruby/jruby/builds/637731773 [163 min 13 sec]
travis-ci has left #jruby [#jruby]
* enebo[m] uploaded an image: IMG_20200115_193745.jpg (3141KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/XPajXBlydQLywEfNQFalKRki >
<enebo[m]> rwilliams I recently inherited the rug and have not fully put it in place yet
<rwilliams[m]> enebo: :) What is the resolution on that monitor?
<enebo[m]> 3440x1440
<enebo[m]> Also 144Hz
<rwilliams[m]> Nice. I've got monitor arms for mine so i can swivel one vertical for an editor but my desk is currently too close to the wall for arms
<enebo[m]> yeah this was a compromise between have a lot of real estate for work and reasonable refresh
<enebo[m]> toughest part was getting the right KVM so I could share it and still pass through the 144Hz
<rwilliams[m]> Yeah
<enebo[m]> but one 34" is enough real estate for me. Some day we will have 8k megadisplays
<enebo[m]> You will be able to see hair follicles if you look close enough
<rwilliams[m]> I wish apple would re-release an upgraded version of their original thunderbolt display. The pro display xdr is just silly. 1000 bucks for a freaking stand?
<enebo[m]> Man I really need to straighten up those cables
<enebo[m]> hahah yeah
<enebo[m]> I am largely done with apple unless something big changes
<enebo[m]> My only regret moving off is keynote but I still have a 2011 MBP which is still somewhat working for that
<rwilliams[m]> I bought the 2016 macbook pro upgrading from a 2013 macbook pro and it was the worst purchase i've ever made computer related imo.
<enebo[m]> I think that keyboard is a firable offense, but it is not just that
<enebo[m]> magsafe removed. the way oversized track pad. no escape key. taking waaaaay to long to change intel generations
<enebo[m]> I have a feeling loss of 32 bit will piss some people off
<enebo[m]> I half feel sympathy for Apple on that. You gotta kill support at some point
<rwilliams[m]> All the steam folks who lost significants games from their library are for sure pissed
<enebo[m]> yeah I bet
<enebo[m]> Anyways. I like their case and Keynote at this point
<rwilliams[m]> I use docker a lot and once I got a taste for native linux docker compared Docker Desktop for mac via AppleHV there is just no going back
<enebo[m]> I really did not have many issues switching back to Linux after a long ride on MacOS
<enebo[m]> I think largely Gnome settings and their UX not providing really basic config is the worst of it
<enebo[m]> Generally you can do a lot a customization but you need to fire of dconf and figure out the magic
<rwilliams[m]> Yeah. I had a lot of little tweaks I had to do on my linux setup and it was some obscure stack overflow stuff. And then I had a power outage and it screwed up grub somehow. Not even sure how that is a thing
<enebo[m]> ouch
<rwilliams[m]> There is a general tool to repair grub but it adds like ever possible boot option, like hard links to efi files and the like. So now i have like 20 boot options because i'm too lazy to remove them
* rwilliams[m] uploaded an image: IMG_20200115_180329.jpg (423KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/uyjiARgDWfKsSJXrCkiBdKiI >
<rwilliams[m]> I may have exaggerated a bit but still this is ugly
<enebo[m]> hahah I am looking for GNU Hurd in that list
ur5us has quit [Ping timeout: 260 seconds]
ur5us has joined #jruby
ur5us has quit [Ping timeout: 260 seconds]
travis-ci has joined #jruby
travis-ci has left #jruby [#jruby]
<travis-ci> jruby/jruby (master:c48536c by Charles Oliver Nutter): The build was canceled. https://travis-ci.org/jruby/jruby/builds/637713551 [178 min 17 sec]
Iambchop has joined #jruby
Iambchop has quit [Ping timeout: 245 seconds]
Iambchop has joined #jruby
den_d has joined #jruby
Liothen has joined #jruby
fidothe has joined #jruby
ur5us has joined #jruby
shellac has joined #jruby
rusk` has quit [Read error: Connection reset by peer]
rusk` has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
ur5us has quit [Ping timeout: 260 seconds]
shellac has joined #jruby
shellac has quit [Ping timeout: 258 seconds]
shellac has joined #jruby
lopex has joined #jruby
lucasb has joined #jruby
abe_ has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
xardion has quit [Remote host closed the connection]
xardion has joined #jruby
shellac has joined #jruby
<rwilliams[m]> If any of you are VS Code folks and you hate how editors suggetions won't allow you to type a new variable/class/method name and hit enter, please upvote this backlog candidate. It only needs 20 upvotes to make the backlog.
<rwilliams[m]> * If any of you are VS Code folks and you hate how editor suggetions won't allow you to type a new variable/class/method name and hit enter, please upvote this backlog candidate. It only needs 20 upvotes to make the backlog.
<rwilliams[m]> I guess i should link it
rusk` has quit [Remote host closed the connection]
shellac has quit [Quit: Computer has gone to sleep.]
shellac has joined #jruby
ur5us has joined #jruby
ur5us has quit [Ping timeout: 260 seconds]
shellac has quit [Quit: Computer has gone to sleep.]
fzakaria[m] has joined #jruby
<fzakaria[m]> Sorry I hope people don't mind me asking a `warbler` question in this room -- although it's under the JRuby organization.
<fzakaria[m]> I've been using warbler to build a gem where one of the dependencies are `:path =>` to a local gem (monorepo)
<fzakaria[m]> Not a big deal but it causes 2x files to be included for size/bloat
<fzakaria[m]> I've noticed that warbler is included the built gem & the source root itself from the `:path`
<fzakaria[m]> (What's ChanServ)
<rwilliams[m]> Did you just get a spam channel invite?
<fzakaria[m]> yea
<fzakaria[m]> (This is my first time btw using Matrix & thru Riot app -- so slow to send a message hmm)
<rwilliams[m]> I get that too some times
<rwilliams[m]> The spam, not the slowness
ur5us has joined #jruby
victori has quit [Ping timeout: 260 seconds]
shellac has joined #jruby
lucasb has quit [Quit: Connection closed for inactivity]
ur5us has quit [Ping timeout: 260 seconds]
shellac has quit [Quit: Computer has gone to sleep.]
ur5us has joined #jruby
abe_ has quit [Remote host closed the connection]
<cshupp[m]> ezzeddine: Last large rails app I upgraded from 1.7 to 9k became very slow too.
<cshupp[m]> Slowness came when entropy got too low.
<cshupp[m]> You can check entropy levels via:
<cshupp[m]> cat /proc/sys/kernel/random/entropy_avail
<cshupp[m]> If entropy is low installing this:
shellac has joined #jruby
<cshupp[m]> will save yo
<cshupp[m]> * will save you.
<rwilliams[m]> good stuff
<cshupp[m]> JRuby now depends on Secure Random, but on quiet linux boxes, secure random is known to block for long periods of time.
<cshupp[m]> Actually 9.0.4 didn't ave the issue, so not all of 9k suffered from the slowness.
<cshupp[m]> * Actually 9.0.4 didn't have the issue, so not all of 9k suffered from the slowness.
<cshupp[m]> I can't recall the number for 'low', but if I recall right the hundreds is low. The thousands is pretty good.
<headius[m]> I thought we'd mostly hacked around that, forcing the jvm to use /dev/urandom
<headius[m]> It's a hard thing to confirm though
<cshupp[m]> I know my old group forgot that trick and I think it was 9.2.6 and the issue went away with haveeged, but I might recall incorrectly.
<cshupp[m]> Is the torqebox server coming back?
<rwilliams[m]> I highly doubt it
<cshupp[m]> sigh, I can't get my build back so far...