<headius>
enebo: so where do you stand on 9.2 support? I was about to say it is EOL but I realized we only released 9.3 a bit over a year ago.
<headius>
this prepend thing came from JetBrains on behalf of users that "can't" upgrade from an older JRuby 9.2 (though the reasons given make me wonder if they could even upgrade to 9.2.x if we release a fix).
<enebo[m]>
yeah I have been thinking about this and I am leaning towards just saying EOL
<headius>
me too
<enebo[m]>
The main problem I have is actual dev cycles to put into our project
<enebo[m]>
We are just putting out a huge step up but we still have a lot left to do
<headius>
the issue he links has a user that can't leave 9.2.8 because there aren't premade Alpine images for 9.2.9 or later, so I'm unsure what if anything we can do to help
<enebo[m]>
yeah
<headius>
we could add Alpine images I suppose
<headius>
but he'd have to update to 9.2.21 to get them
<enebo[m]>
It makes me wonder how you can even use 9.2.9
<enebo[m]>
err 9.2.8
<headius>
yeah
<enebo[m]>
In that sense I am very confused since 9.2.8 was not all that solid and if that is working then why wouldn't any other version
<enebo[m]>
like 9.3
<enebo[m]>
I tend to dislike the "hey just upgrade come back" but with that said we have 21 releases for a reason
<enebo[m]>
Also 9.3 is a better 9.2 in the sense things like prepend do work
<enebo[m]>
or work more
<enebo[m]>
So postgresql is running to the point I get errors but it seems a lot has changed
<enebo[m]>
Also mysql and sqlite3 took a much more copy the same code approach than postgresql
<enebo[m]>
it does copy mostly but it replicates a lot of files
<enebo[m]>
anyways I am trying to get that crud'able
<headius>
sounds great
<headius>
I am triaging for the moment and then will make some tweaks to my talk, but it should mostly be what I presented at RubyConf
<enebo[m]>
I am trying to think of the best way of even doing this
<enebo[m]>
for sqlite3 it tends to be a single file compare
<enebo[m]>
I am thinking I might just dump a massive dump of AR between last 6 and 7 and see what changes
<enebo[m]>
which is going to be a big diff
<enebo[m]>
What I have done so far has just been tweaking until it runs a little more which I think will create some frankenadapter
<headius>
yeah looking at the diff might be simplest
<headius>
it's so frustrating that they don't see value in unifying these adapters better, or using a common backend
<headius>
I'm doing a release of jffi to pick up some contributed arm32 fixes
<lavamind>
yay
<lavamind>
I've also submitted a patch on jffi for mipsel
<lavamind>
I also have a patch for riscv64 in the queue :)
<lavamind>
I'm curious to know what would be the EOL for 9.3, since this version is likely the one going into the next stable version of Debian
<lavamind>
I have 9.4 in my sights but its unclear if it would be compatible with Puppetserver, which is why we're packaging it in the first place
<headius>
ldywicki: hey I forget where we left off with dockcross work, but I have a need for it now: mips32/mipsel builds
<headius>
the GCC compile farm does not have such a machine
<headius>
neat, debian folks will get us temporary access to their build infra
<headius>
so that's good enough for now
<lavamind>
headius: I'm wondering if you can see my messages at all in here :p
<lavamind>
I'm @jcharaoui on github
<headius>
enebo: this guy reports he's on #jruby on libera but messages are not coming through to here
<headius>
IRC integration may have broken
<headius>
last connection I see in the bridge logs was in June
<enebo[m]>
there is some message about no end to end enryption in bridge status
<enebo[m]>
I am not sure that is really related though
JrmeCharaoui[m] has joined #jruby
<JrmeCharaoui[m]>
hello!
<headius>
oh hey there!
<headius>
you found us!
<headius>
we'll look into the libera bridge, must have lost connection somehow
<JrmeCharaoui[m]>
so that would explain the very old /topic on IRC haha
<JrmeCharaoui[m]>
I'm seeing your messages there
<enebo[m]>
headius: did you change something for the bridge to work?
<headius>
oh so it's not working both ways
<headius>
enebo: it has been a long time since I did anything with it
<enebo[m]>
ok
enebo has joined #jruby
<JrmeCharaoui[m]>
I just thought you all must have had agressive filtering on messages from randos like me lol
<enebo>
gmm
<headius>
ok so mipsel
<JrmeCharaoui[m]>
so, I guess I should introduce myself
<enebo[m]>
whot...it shows me joining
<headius>
enebo: but no messages coming through
<JrmeCharaoui[m]>
for several weeks I've been working on updating the jruby package in Debian
<headius>
yeah that's great, thanks for the help
<JrmeCharaoui[m]>
I've uploaded 9.3.9.0 a few days ago
<enebo[m]>
yeah matrix -> is showing and join status on libera
<JrmeCharaoui[m]>
as part of that work I've also had to update the jffi package
<JrmeCharaoui[m]>
and because no other reason that I like making this hard for myself 😅 I've enabled the testsuite on jffi as part of the build process
<enebo[m]>
appservice: hello my dude
<JrmeCharaoui[m]>
thats what led me to look into 32-bit arm and mispsel
<headius>
got it... yeah we don't have a good way to test those so they either regressed or never fully worked in the first place
<JrmeCharaoui[m]>
I understand you'd like to have access to a mips machine to generate the binary
<headius>
yes please! We have a user that is helping us set up some dockcross builds, but at the moment we have no way to make mipsel binaries
<JrmeCharaoui[m]>
I've only started working with "porter boxes" as we call them but there is a procedure to grant guest access
<JrmeCharaoui[m]>
there are several ones related to the default java class
<JrmeCharaoui[m]>
I'm pretty sure they have a related cause
<JrmeCharaoui[m]>
I don't know enough about jruby to figure out if that's a serious problem
<JrmeCharaoui[m]>
eg. Error: test_compareTo_non_comparables(TestComparable): ArgumentError: Java package 'default.package.class' does not have a method compareTo' with 2 arguments`
<JrmeCharaoui[m]>
and Error: test_extend_default_package_class(TestHigherJavasupport): TypeError: superclass must be a Class (Java::JavaPackage given)
<JrmeCharaoui[m]>
there's really no rush to figure it all out
<headius>
yeah I see them
<headius>
or I see some of them
<headius>
JRuby 9.3.9 I see
<JrmeCharaoui[m]>
yes, the release was made as I was putting the finishing touches on the package 😅
<headius>
going to try to run these locally on 9.3.9 and Java 17
<headius>
it's possible it's a java 17 issue in 9.3 that has not been fixed
<JrmeCharaoui[m]>
aah ok perfect
<JrmeCharaoui[m]>
wow, thank you so much
<JrmeCharaoui[m]>
I'm reluctant to package 9.4 immediately because the main reason we're going it is for puppetserver, and it hasn't declared itself compatible with 9.4 yet (understandably)
<headius>
sure that's reasonable
<headius>
it's barely been out a week
<JrmeCharaoui[m]>
personally I'm comfortable shipping 9.3 in the next version of Debian stable, especially if you are as well
<JrmeCharaoui[m]>
the freeze is happening in like two months so that doesn't leave a lot of time
<headius>
we'd certainly like to have 9.4 out there since it is a big leap forward on compatibility
<headius>
but I understand if the timing is difficult
<JrmeCharaoui[m]>
yeah I know, Debian has nearly finished its own transition to ruby 3.1 by default
<JrmeCharaoui[m]>
it's possible puppetserver would run just fine with jruby 9.4, we'll have to try it
<headius>
yeah that would be very nice
<JrmeCharaoui[m]>
ok so to recap, I'll submit PRs for riscv64 and the jni makefile
<JrmeCharaoui[m]>
and to strip out -m<MODEL>
<headius>
sounds good, I will try to reproduce these failures
<JrmeCharaoui[m]>
awesome!
<headius>
if we don't need MODEL at all those lines should just be deleted
<JrmeCharaoui[m]>
I'm really happy to have made this connection btw, so thank you for your support
<headius>
yeah no problem, sorry the IRC stuff was not working right!
<headius>
thank you for the help
<JrmeCharaoui[m]>
hehe yeah hopefully the bridge will get fixed at somne point
<JrmeCharaoui[m]>
np, ttyl
<headius>
ok
enebo has left #jruby [#jruby]
<headius>
ok test:jruby just hangs on aarch64 (M1)
<headius>
so that doesn't help much but I'll try the individual tests'
subbu has quit [Quit: Leaving]
<headius>
spec:ji is green locally
<headius>
not sure where these linux aarch64 failures are coming from
<headius>
(on debian build server)
<headius>
Jérôme Charaoui: I have not been able to reproduce these so far on MacOS but I'll try on Linux when I get back to my machine
<headius>
spec:ji is green on 9.3.9 + java 17; test:jruby hangs but does not appear to have the same errors up to that point