trying to refresh myself on the discussion. The problem is I dont want to add a way to add aliases directly to the registry to the modders
#1.20 Porting
kind stirrup
yes i've given up on having worlds compatible between MC versions
Modders fuck that concept and ive gotten tired of trying to teach them otherwise
hexed stratus
It's ready I need the RB to be built so I can do changelog stuff
paper hedge
you want to rip out that system after it was just added as non-public API?
kind stirrup
No. I don't want it public api, which is why it was added in the way it was.
paper hedge
excuse me?
kind stirrup
... really.
dusty spruce
... really.
Sorry that post I just sent bothered you. Deleted it right on cue.
flat zealot
gah, I always forget, what was the new gradle 8 replacement for archivesBaseName
kind stirrup
RIGHT I remember the status of the alias stuff
we were supposed to design a API in DR for it
not just 'make the method public'
paper hedge
gah, I always forget, what was the new gradle 8 replacement for `archivesBaseNam...
are you sure that's gone?
paper hedge
we were supposed to design a API in DR for it
Yeah we were discussing it last night
kind stirrup
for now, mark it as 1.20 so the depreication checker shuts up.
paper hedge
.
kind stirrup
doesnt amtter much as its a private method. But ya
flat zealot
well this is fun, the forgespi dep doesn't seem to pull transitives?
kind stirrup
i would be more inclined to do register(Name, Supplier, Set<ResourceLocation> aliases)
DR isnt a builder system
flat zealot
it's definitely in the pom
paper hedge
i would be more inclined to do `register(Name, Supplier, Set<ResourceLocation> a...
sure, simple enough
I was concerned that was going to exponential inflate the # of register methods, but there's actually only one register method in DR right now
What there is a lot of is static factories for DRs and ROs
kind stirrup
Ya.
That is the concern of doing this vs a builder
Techically this doubles the number of methods 😛
teal plover
well this is fun, the forgespi dep doesn't seem to pull transitives?
is this the FML subproject?
kind stirrup
However, i think that this is something that hasnt been touched in a long enough time that we dont have to worry about the potential future 'what if we add more arguments'
flat zealot
yeah I figured
hushed sapphire
yeah, you have to manually add all transitives that you need. because the installer requires a flat list.
flat zealot
> Task :forge:validateDeprecations FAILED
21 actionable tasks: 4 executed, 17 up-to-date
Deprecated method net/minecraftforge/common/ForgeMod#addAlias(Lnet/minecraftforge/registries/IForgeRegistry;Lnet/minecraft/resources/ResourceLocation;Lnet/minecraft/resources/ResourceLocation;)V is marked for removal in 1.20 but is not yet removed
FAILURE: Build failed with an exception.
kind of annoying that it prevents me from getting patches
kind stirrup
for now, mark it as 1.20 so the depreication checker shuts up.
.
flat zealot
I'm working on that rn
on what
hushed sapphire
question. line in changelog: Forge dummy entries handled by Forge registries have been completely removed.
what is that referring to?
paper hedge
kind of annoying that it prevents me from getting patches
that's kinda the point
hushed sapphire
what does this mean too: The Forge common config has been removed from the game itself
flat zealot
git is telling me that no patches have changed 
flat zealot
what does this mean too: `The Forge common config has been removed from the game...
the common config was deprecated, it was only previously used by resource caching
now it's an empty file, as before
paper hedge
apparently we need to restore the dummy system in ForgeRegistry because it was improperly deprecated for removal
flat zealot
can the installer handle classifiers?
paper hedge
yeah there's code in there to specialcase them
hushed sapphire
apparently we need to restore the dummy system in ForgeRegistry because it was i...
probably not. it was deprecated by lex's choice. i think it's still useful in certain scenarios, but lex disagrees.
paper hedge
That would probably be good
flat zealot
this is actually harder than expected because the deprecation and todo commits had vanilla sources
paper hedge
oof
hushed sapphire
Ouch
dusty spruce
Eesh
flat zealot
yeah I'd be screwing it more
paper hedge
this is actually harder than expected because the deprecation and todo commits h...
revert the sources removal commit then and try again 
flat zealot
it doesn't really work though
like if we want to squash those commits, those commits need not to contain the vanilla sources
because the removal of sources would be squashed in the first commit
paper hedge
the point is to put them on top of the squash. you can fix them by genning patches and not commiting the changes to the source
flat zealot
ah it gets even better.. there's conflicts with the commits after.. yeah I don't see this doable now, and since the primer documents all changes, we can do the split commits next time
paper hedge
@eager roost https://github.com/MinecraftForge/MinecraftForge/commit/3de22ffba3c5325f29c39b66c0c8a02e2d86f9e4 so do we just change the LOGGER.fatal to throw?
frail flare
My intention was to do both the logging and the throw
flat zealot
lol bad shrimp
paper hedge
actually it didn't
flat zealot
gah, I always forget, what was the new gradle 8 replacement for `archivesBaseNam...
^ btw
paper hedge
I just assumed "Dennis C" was giga for some reason
paper hedge
My intention was to do both the logging and the throw
k
frail flare
I just assumed "Dennis C" was giga for some reason
Nah, that's me 🤣
paper hedge
you know that probably means I have attributed more commits to giga in my head in the past
¯_(ツ)_/¯
frail flare
~~it sounds very german though, yes~~
It's not even the German spelling though, that would be with a single n for some reason 😄
flat zealot
I by now can associate the name of most people that commited to forge to their online usernames
don't ask me why
for the readme, what's the new LTS?
paper hedge
could remove it
flat zealot
anything remaining for the update?
plush current
Regarding this, did you set the project name in the settings.gradle to fix it? <#neoforge-private message>
flat zealot
fix what
frail flare
anything remaining for the update?
Nothing pressing, but that lighting thing in ChunkAccess is still open
paper hedge
what thing
frail flare
^
frail flare
I have an idea for it but it's kinda ugly
plush current
fix what
I'm going by the commit message here. Assuming you changed the build.gradle to the non-deprecated one for setting the archiveBaseName. Ideally you would get rid of that entirely and set the project name correctly in the settings.gradle
flat zealot
@barren snow RB is out, fix ya blog post 
(sorry)
frail flare
Wrong one
paper hedge
lol
paper hedge
- [ ] Update ForgeGui to match Gui changes did this get checked at all
is there any new gui features this update?
in 1.17 with powdered snow they added the frozen hearts and frozen overlay and we broke that
flat zealot
is there any new gui features this update?
no
flat zealot
yeah I did see that
flat zealot
baseless claims.. /hj
kind stirrup
manifest caching/delay issue?
flat zealot
we need to decide on the next LTS
paper hedge
I updated the description of the TC project cuz it was wrong and it prompted a rebuild
paper hedge
we need to decide on the next LTS
Curle wanted to possibly involve the community
therefore I say we remove the table entry for release
it can be added back later after a decision
livid sequoia
Fair enough, my thought would be 1.19, though I don't know whether .2 or .4
livid sequoia
give the retrogradle team more work why don't'cha
undone tulip
if 1.19 probably 2 because many big mods didn't update past that
plush current
then again that was only semi-supported vs 1.19.4...
paper hedge
It's running now
I don't understand why it didn't work earlier. It ran the moment the version was published (9:59 am CST my time) -- the teamcity start time matches this. So perhaps it ran too fast?
And the snapshot one didn't run at all
livid sequoia
Maybe put a minute delay?
paper hedge
@austere surge this would be your area of expertise. Also, the snapshot snowblower hook should still run on releases as that is part of the git history
flat zealot
then again that was only semi-supported vs 1.19.4...
most mods have skipped 1.19.4
livid sequoia
Makes sense when you look at what mojang did in .3 with the datagen not in code
flat zealot
have you updated deps shrimp?
austere surge
<@304594425286361088> this would be your area of expertise. Also, the snapshot s...
Trigger can be broken
flat zealot
modules are not happening in the initial release
paper hedge
austere surge
I will check tomorrow morning.
austere surge
Well that is weird
paper hedge
Huh?? https://i.imgur.com/mHJOje6.png
why is it saying 1.20 released 5 days ago
flat zealot
and the plat types rfc is interesting and still needs discussion
austere surge
Since the trigger uses the same manifest
paper hedge
¯_(ツ)_/¯
hexed stratus
<@113392954206347264>
I know doing some formatting is all
paper hedge
wait wtf
mojang messed up the release time for 1.20
it's june 2nd in their manifest data
haughty phoenix
why is it saying 1.20 released 5 days ago
Same for the mcage command, moja messed up
flat zealot
lol
kind stirrup
That would be the only reason it would say that on the git
next maple
That makes it hard to port forge on the same day as the release 😃
undone tulip
they did it intentional to make you look bad
flat zealot
yeah we were only done yesterday
this is not right!
@signal lion speaking of blog posts, may I join the blog team I also need to update my optifine post because I've realised I've linked the fabric version of a mod, not the forge one
signal lion
Yes you may
flat zealot
that response is really giving me "teacher, may I go to the toilet" flashbacks
kind stirrup
Why didn't you go between classes!
flat zealot
I was busy working on waifu, mmmkay
anywho, bathroom breaks aside... anything left on the todo?
livid sequoia
Did shrimp finish the alias thing? If so, it's just removals and renames after push
paper hedge
lex said change it to 1.20 deprecation
probably so it can be fixed separate to the squash commit
I'm currently touching forgegui using snowman as a reference cuz there are a few desyncs but nothing major
kind stirrup
There is nothing to 'fix' about the alias thing
Its a API that needs to be designed and addressed later
the deprecation is on a private method as a reminder, which evedently got forgotten
livid sequoia
Then, I don't think there's anything left other than changes after squash
paper hedge
well
I just said I'm fixing a few things in forgegui
it's my last todo bullet point that i use ever update
livid sequoia
Oh, I thought there was a desync in snowman, misread
paper hedge
wow running forge is super fast now
without the double compile and without invoking gradle
paper hedge
change to intellij then
livid sequoia
PSS?
signal lion
ProjectScopeServices
livid sequoia
Ah
flat zealot
oh curle can you approve my join request. https://github.com/orgs/MinecraftForge/teams/bloggers
* What went wrong:
Could not determine the dependencies of task ':forge:filterJarNew'.
> Joined SRG jar not found
great
Error getting artifact: net.minecraft:joined:1.20-20230608.053357:srg@jar from MCPRepo
java.net.SocketException: Connection reset
paper hedge
btw I never copied this over https://github.com/MinecraftForge/MinecraftForge-Private/commit/654828be0a7f91fcdab0f9ece755e441dd02ab04
flat zealot
lol
dusty spruce
oh froge not forge
He means that Forge is on hold.
hexed stratus
oh froge not forge
can we rename it it kicks my dyslexis right in the teeth :P
worn bison
tode?
kind stirrup
yes we need to rename it, if it ever becomes a production system
paper hedge
so I think a squash is in order
well we should rebase on the 1.19.4 RB commit
if maty/whoever wants to do the squash I will let them otherwise I can do it; just needs to have the proper info as a highlight summary of all of our commits + all the co-authors
livid sequoia
well we should rebase on the 1.19.4 RB commit
There was one addition, so probably worthwhile
flat zealot
if maty/whoever wants to do the squash I will let them otherwise I can do it; ju...
oh yes I've been waiting to do a squash for a while
livid sequoia
Wait no, that was 1.19.2
flat zealot
should I rebase or cherry pick
paper hedge
rebase first I think would be cleaner
flat zealot
though actually it needs to be a rebase, nvm
paper hedge
don't squash until you push the commit updating the readme
you can probably force push the rebase to 1.20-dev
we don't really use the rebase/1.x thing and it doesn't fit the branch scheme, although it's all a suggestion. but generally it should be 1.xx.xx-(dev|rebased|final)
livid sequoia
yeah dunno why champ did it this way
I started to organize my branch in folders, so I just sorta went, yeah, we would probably rebase at random points in time, so let's just make a timestamp
flat zealot
well fair enough ig
ah yes, intellij is throwing up because it's attempting to reload gradle after every rebased commit
livid sequoia
cli ftw
flat zealot
stabby
paper hedge
it threw up on the first commit and that was it
for me at least
go ahead and do the squash if you want and put it on 1.20-final
I forcepushed to 1.20-dev shh
flat zealot
when you pushed that my rebase was done lol
paper hedge
xP
hexed stratus
<@113392954206347264>
it is done
paper hedge
what happened
flat zealot
#neoforge-private I somehow managed to push my branch instead of pulling
thanks gh desktop
paper hedge
what did you do
flat zealot
I checked out the branch with gh desktop and instead of promopting me to pull it promopted me to push
hexed stratus
do you need me to do some gitfu and split it into a branch?
hexed stratus
everyone is screwing up here :P
paper hedge
pushed the wrong local branch 
my prior force push just deleted commits
infact it was the 32 commits that you pushed
paper hedge
the folder strategy could be nice
at least in intellij it sorts by folder
so like 1.20/dev 1.20/final? idk
u r squashing rn yes @flat zealot? also push the readme changes
paper hedge
no I thought you were
flat zealot
lol
it's fine
give me one extra second
thonk, I spent 30 seconds looking for the readme, why's it in docs/ 
paper hedge
idk but github finds it all the same
paper hedge
if you're not going to contribute anything useful please go to other channels thanks 👍 this isn't for normal users, it's our channel for development that happens to be public so that the public can know what's going on
undone tulip
maybe you should check if anything got lost during the rebase and push mistakes
paper hedge
?
undone tulip
if your force push removed a commit that maty didn't push afterwards
paper hedge
I force pushed the wrong local branch, maty's client recognized it cuz the history was still matching their local for 1.20-dev, and then I fixed it afterwards
nothing is missing
paper hedge
maty is squashing, and hopefully writing a good commit message attributing all co-authors, and then there is testing to do
undone tulip
if you need help testing I'm here
paper hedge
we'll see. I assume we will give an installer jar in squirrels at some point; we've done that in the past
flat zealot
- Creative mode tabs are now a registry; renamed the
BuildContentsevent toBuildCreativeModeTabContentsEventand moved it in its own class ScreenUtilswas removed in favour of aGuiGraphicsextension- Update to Gradle 8 and FG6
- Rename
SaplingGrowTreeEvent->BlockGrowFeatureEvent - Remove the Forge common config
- Remove registry dummy entries
- Fix
RemappingVertexPipelinenot forwardingendVertex()call
hmm, I feel like that's the highlights
paper hedge
screenutils was not removed
ash reverted some of his removals/changes
and it's still SaplingGrowTreeEvent
flat zealot
I thought I reverted his revert
paper hedge
no don't
flat zealot
- Creative mode tabs are now a registry; renamed the
BuildContentsevent toBuildCreativeModeTabContentsEventand moved it in its own class ScreenUtilswas deprecated in favour of aGuiGraphicsextension- Update to Gradle 8 and FG6
- Remove the Forge common config
- Remove registry dummy entries
- Fix
RemappingVertexPipelinenot forwardingendVertex()call - Remove Forge tool tags in favour of vanilla ones
paper hedge
what was in the forge common config
Update to Gradle 8 and FG6 - this could mention the MDK update
flat zealot
the old stuff for resource caching
paper hedge
ah maybe mention that cuz it's confusing, like it had no more properties so it was removed for the time being
And also maybe mention the pack format is now 15 for both server and client resources
frail flare
Do resource pack and data pack version actually match again?
paper hedge
Yep
flat zealot
when was resource caching removed?
paper hedge
Remove the Forge common config - I'm just weary to word it this way as people will take it as "OH NO COMMON SYSTEM IS GONE ENTIRELY!!!"
flat zealot
can someone go find out
paper hedge
why
flat zealot
- Creative mode tabs are now a registry; renamed the
BuildContentsevent toBuildCreativeModeTabContentsEventand moved it in its own class ScreenUtilswas deprecated in favour of aGuiGraphicsextension- Update to Gradle 8 and FG6
- Remove the Forge common config file (it only contained the deprecated old fields for resource caching, which was removed in 1.19.3)
- Remove registry dummy entries
- Fix
RemappingVertexPipelinenot forwardingendVertex()call - Remove Forge tool tags in favour of vanilla ones
- The pack format is now 15 for both resource packs and data packs
better?
paper hedge
the 2nd bullet point is inconsistent in wording
and the pack format bullet point could go closer to the top cuz the bottom bullet points all seem like "fix" or "remove"
flat zealot
I wonder whether I should use passive for all bullet points
paper hedge
that's usually how commits are written
flat zealot
- Creative mode tabs are now a registry; the
BuildContentsevent was renamed toBuildCreativeModeTabContentsEventand moved it to its own class - The pack format is now 15 for both resource packs and data packs
ScreenUtilswas deprecated in favour of aGuiGraphicsextension- Forge and the MDK were updated to Gradle 8 and FG6
- The Forge common config file was removed (it only contained the deprecated old fields for resource caching, which was removed in 1.19.3)
- Registry dummy entries were removed
RemappingVertexPipelinewas fixed to forward theendVertex()call- Forge tool tags were removed in favour of vanilla ones
flat zealot
this is everyone, right?
flat zealot
what did you edit?
missed a the in second to last
paper hedge
looks like it is everyone
push to 1.20-final
what was just deleted I wasn't paying attention? :o
nvm apparently nothing was my eyes deceive me
paper hedge
- Creative mode tabs are now a registry; the `BuildContents` event was renamed t...
wait
"favor"
we use american english
paper hedge
did you review my MDK changes btw
I think it should be good but
Also we have been calling the main commit msgs Forge 1.xx since 1.19
so maybe you want to go for that too
flat zealot
how do I exit the ammend thing
paper hedge
idk
lol
hmm
the git ignore was changed to allow committing rejects, just now noticed, I guess that's fine?
flat zealot
yeah
paper hedge
Ok time to build an installer and test things and give it to squirrels
flat zealot
applyPatches is also verbose
thought arguably it should've been verbose from the start
paper hedge
oh that got kept? that's fine yeah
flat zealot
what's this prelaunch thing
paper hedge
Huh?
flat zealot
paper hedge
Oh
yeah that was a thing I did we discussed it I don't fully remember
basically the hashmap ordering in jpms/sjh/somewhere was different in forgedev and caused the modlauncher log4j2.xml to be loaded instead of fmlloader's
so it subtly changed the way logging looks
flat zealot
paper hedge
I saw those
flat zealot
you know what's now missing from the mdk. mixingradle
paper hedge
let's not have that discussion rn
flat zealot
a quick look over the patches and nothing stands out, so I'm going to shut down my laptop and go to bed
paper hedge
Oh ok bye thanks for ur work!
I'll compile the installer then
(well, I already was)
pearl light
nothing changed that would mean the MDK wouldn't be compiling, yes?
paper hedge
?
flat zealot
yes something changed and the mdk wouldn't compile (tabs). but I fixed that
pearl light
good good
pearl light
yeah, we still don't have the necessary configuration to test that with a task
flat zealot
Did you test it compiled
I wrote it in the forge sourceset
paper hedge
ok then
flat zealot
I wanted to add it as a sourceset or something and add a configuration to launch a mdk client, but that's more work
pearl light
also, just to confirm, there's no errors in the test sourceset either? since our tasks don't poke that much either (outside of test or the test run configs)
paper hedge
I'll double check
paper hedge
#squirrels-🦊 message
pearl light
we should really make a porting and new version checklist at some point
paper hedge
yeah true
flat zealot
I hope it works fine, I don't want to lose my squash 
paper hedge
I'll double check
my intellij is erroring
this is all wrong https://i.imgur.com/hLyKhz4.png
dusty spruce
I hope it works fine, I don't want to lose my squash <:Kekw:1017154416484298792>
Me too.
flat zealot
pearl light
iirc, Adoptium makes a tracking issue with some sort of copy-pasted template text that has a load of checklists and such
and because people who work on it have write access to the repo, they can just check stuff off as they go along
flat zealot
this is all wrong https://i.imgur.com/hLyKhz4.png
didn't you manage to build an installer?
pearl light
thonks to thonks
paper hedge
yeah thru gradle
flat zealot
paper hedge
I deleted the out folder, trying again
flat zealot
your intellij is fucked
undone tulip
installer borked
kind stirrup
the git ignore was changed to allow committing rejects, just now noticed, I *gue...
we really need a "convertToUpdateRepo" and reverse task in the gradle. to automate some of that
subtle dirge
installer borked
reproducible by me as well
flat zealot
installer borked
how is srgutils even there
subtle dirge
want logs?
paper hedge
Yeah that makes no sense cuz srgutils shouldn't be an installer lib
ah
"jar": "net.minecraftforge:installertools:1.3.0",
"classpath": [
"net.md-5:SpecialSource:1.11.0",
"net.sf.jopt-simple:jopt-simple:5.0.4",
"com.google.code.gson:gson:2.8.7",
"de.siegmar:fastcsv:2.0.0",
"net.minecraftforge:srgutils:0.4.11",
"org.ow2.asm:asm-commons:9.3",
"com.google.guava:guava:20.0",
"com.opencsv:opencsv:4.4",
"org.ow2.asm:asm-analysis:9.3",
"org.ow2.asm:asm-tree:9.3",
"org.ow2.asm:asm:9.3",
"org.apache.commons:commons-text:1.3",
"org.apache.commons:commons-lang3:3.8.1",
"commons-beanutils:commons-beanutils:1.9.3",
"org.apache.commons:commons-collections4:4.2",
"commons-logging:commons-logging:1.2",
"commons-collections:commons-collections:3.2.2"
],```But also, 0.4.11 exists
paper hedge
aren't transitives disabled?
not in this case :>
flat zealot
so this is expected behaviour?
paper hedge
oh I think I have a local version of srgutils 0.4.11
so it desynced the checksum
this is why CI exists
paper hedge
flat zealot
and if anything breaks, you can blame it on me because that's how git blame works
undone tulip
moment
deft ridge
we what
flat zealot
lol what
silk garden
flat zealot
how did you just realise now
paper hedge
Because
flat zealot
and why did no-one notice it before
silk garden
Which version do you want?
Yes
flat zealot
I want all of them!
meager cape
the more versions the better
silk garden
so never?
pearl light
ASM 1.0 or bust
undone tulip
I'm just gonna silently ignore that cuz we already squashed <:thinkies:932605814...
you could do another squash of the fixup commit
paper hedge
I could
flat zealot
but then he can't blame me
paper hedge
but this requires updates to our libs or a change in how we declare classpath deps in the installer profile
flat zealot
and he would love to blame me
imagine a happy shrimp blaming me for anything that went wrong during porting
paper hedge
yes, git blame
flat zealot
shared blame! team work!
undone tulip
I think this should be fixed before forge 1.20 is published
flat zealot
too bad git is totalitarian
pearl light
the team that works together, gets blamed together
paper hedge
it's a minor issue, it just means bloat on people's pcs. Also, it's already been an issue for a while
paper hedge
#squirrels-🦊 message
so uh
maty
are you the one that did this mergetool shit?
it broke
flat zealot
define broke
paper hedge
[16:31:48] [main/FATAL] [cp.mo.mo.LaunchPluginHandler/MODLAUNCHER]: Encountered serious error loading launch plugin service. Things will not work well java.util.ServiceConfigurationError: cpw.mods.modlauncher.serviceapi.ILaunchPluginService: Provider net.minecraftforge.fml.loading.RuntimeDistCleaner could not be instantiated
Caused by: java.lang.NoClassDefFoundError: net/minecraftforge/api/distmarker/OnlyIn at net.minecraftforge.fml.loading.RuntimeDistCleaner.<clinit>(RuntimeDistCleaner.java:42) ~[fmlloader-1.20-45.1.1-1.20-final.jar%2367!/:1.0] {}
pearl light
paper hedge
[16:31:48] [main/ERROR] [cp.mo.mo.TransformationServiceDecorator/MODLAUNCHER]: Service failed to load fml cpw.mods.modlauncher.api.IncompatibleEnvironmentException: Missing DistCleaner, cannot run!
[16:31:48] [main/ERROR] [ne.mi.fm.lo.FMLLoader/CORE]: Dist Cleaner is missing, we need this to run
There are infact a lot of errors to check for this
lol
flat zealot
it's in the installer
deft ridge
installer is kill 
pearl light
why is OnlyIn missing
paper hedge
I present to you: modules!
flat zealot
mergetool should've been downloaded and put on the CP
paper hedge
ohno
flat zealot
``java.lang.ClassNotFoundException: net.minecraft.item.ItemGroup``
are you joking
meager cape
paper hedge
the fix here is to just shipit
how wrong these words are so quickly
meager cape
Error occurred initializing CoreMod
java.io.FileNotFoundException: null
I also get this
paper hedge
pearl light
you love to see it
deft ridge
very bad things
pearl light
he performed the noble act of lying /jk
paper hedge
you said the installer can do classifiers!
it does
undone tulip
launching works
pearl light
paper hedge
``java.lang.ClassNotFoundException: net.minecraft.item.ItemGroup``
that's probably cuz you have mods installed
paper hedge
[16:38:00] [main/ERROR] [ne.mi.co.CoreModEngine/COREMOD]: Error occurred initializing CoreMod
16:38:00.083
java.io.FileNotFoundException: null``` this part is real thosomeone deleted intrinsic_tag_appender_binary_compat without removing it from the json
@flat zealot hey u wanna resquash?
flat zealot
yeah I can, but is it actually going to work
paper hedge
btw it was maty who did that unironically
flat zealot
and what's with mergetool and servers
pearl light
how did that not error out in forgedev
paper hedge
idfk
paper hedge
how did that not error out in forgedev
you know it probably did
and we just didn't see it
well yeah we need to wait to resquash until we fix mergetool too
undone tulip
creating new world works
pearl light
:catjam:
west lark
pearl light
shipit time
livid sequoia
Whait, is that the server working?
pearl light
if someone wants to run the server, they can just install the client and invoke the server main class with the client command line /jk
flat zealot
can't shipit unless server is fixed
livid sequoia
What's the issue on the server?
meager cape
Downloading library from https://piston-data.mojang.com/v1/objects/15c777e2cfe0556eef19aab534b186c0c6f277e1/server.jar
java.net.SocketTimeoutException: Read timed out
I'll install a server and test
flat zealot
something with mergetool
livid sequoia
Oh, that fun
paper hedge
I'll install a server and test
it'll break immediately
livid sequoia
Why does mergetool and forgespi have the same distmarker package anyways?
flat zealot
how on earth did it break is the question
it did, now it doesn't
now it works on clients
but not on servers
a "wth" moment
pearl light
which one has the distmarker package?
flat zealot
mergetool
pearl light
and is mergetool api/distmarker included in the set of libraries we pack into the installer?
paper hedge
it is definitely not
flat zealot
have you checked if mergetool is on the classpath
pearl light
hmmmm
flat zealot
wth
paper hedge
wait
fucking hell fuck
D:\Minecraft Servers\Forge1.20Testing\libraries\net\minecraftforge\mergetool\1.1.5-api
vs proper:
D:\Minecraft Servers\Forge1.20Testing\libraries\net\minecraftforge\mergetool\1.1.5
paper hedge
libraries/net/minecraftforge/mergetool/1.1.5/mergetool-1.1.5-api.jar
this is what is in win_args.txt
flat zealot
and where is it downloaded?
pearl light
Forge libraries are downloaded by the installer, iirc
undone tulip
server fails to start https://pastebin.com/JSAYn3Nx
pearl light
so, if I understand this right, the installer has its own copy of Artifact (from FG)
and instead of using the artifact path as encoded in the profile JSON, it constructs the path based on the artifact string (Artifact#getPath()), which means net.minecraftforge:mergetool:1.1.5-api converts to [...]/1.1.5-api
because of the -api part
flat zealot
so the json is wrong
pearl light
were the artifact string net.minecraftforge:mergetool:1.1.5:api -- with the classifier set apart appropriately with : rather than - -- then Artifact would construct the same path as is stored in the version JSON
flat zealot
wait but if there are 3 : in the artifact name, would it properly parse it?
@paper hedge r u taking notes?
paper hedge
I'm confused
"path": "net/minecraftforge/mergetool/1.1.5/mergetool-1.1.5-api.jar"
this is in our version.json that the launcher loads
and it finds it
flat zealot
that TODO in util is what is breaking it
pearl light
if I read the MC 1.20 JSON right, it already has library names which have three : -- for LWJGL and netty-transport artifacts
flat zealot
No that todo is necessary
but it breaks the server
paper hedge
we need to reinvestigate cuz it was still broken the last time and that's not a perfect indicator
pearl light
flat zealot
what if you.. test it
I can do a new squash without the todo.. and if it works everyone is happy
what about that
flat zealot
then test it
kind stirrup
Back, what did yall break?
livid sequoia
The server
kind stirrup
what about it?
livid sequoia
It seems mergetool is being incorrectly referenced such that the runtime cleaner fails when trying to reference any of this classes
kind stirrup
classified artifacts are instead convrted with a version suffix
because the vanilla launcher is/was broken
and didnt support classifiers
But the installer should be fine with it as we encode the actual url
we've been doing this for ages for all the other classified things
flat zealot
so, if I understand this right, the installer has its own copy of `Artifact` (fr...
the artifact name change causes the expected path to change
but the path doesn't change
which breaks the server
flat zealot
we've been doing this for ages for all the other classified things
there doesn't seem to be anything else with classifiers.. at least not on servers
it might work fine in the launcher, but it breaks servers
the good thing is that the client seems to work
paper hedge
I guess we should test whether we need this fix anymore
kind stirrup
the server should be fine
as it has the file path hardcoded
so... whats the issue?
kind stirrup
seems like a simple change, but also dont we already use classified artifacts?
why is this one special?
flat zealot
not on the server
undone tulip
ok so the ender dragon is broken too (no error in log)
it is stuck at its spawn position over the gateway and is not hitable
flat zealot
https://github.com/MinecraftForge/Installer/blob/2.0/src/main/java/net/minecraftforge/installer/json/Artifact.java#L70 and yeah it uses the version in the dir name, which "wrongly" contains the classifier
signal lion
If I had to guess, the ender dragon is an issue with updating multiparts
kind stirrup
<https://github.com/MinecraftForge/Installer/blob/2.0/src/main/java/net/minecraf...
Its not wrong, its intentionally working around a bug in the vanilla launcher
undone tulip
more info on the dragon issue: the dragon is probably moving on the serverside but stuck on the client
pearl light
I have a heavy heavy feeling the problem is in DownloadUtils
since that does workaround the classifier issue, but only for libraries downloaded from the Mojang libraries maven: https://github.com/MinecraftForge/Installer/blob/389603193163d1d9416b0b7e7affdd5c807235e6/src/main/java/net/minecraftforge/installer/DownloadUtils.java#L175C1-L180
since that's the only time it uses download.getPath()
otherwise, the target output is the artifact.getLocalPath() from https://github.com/MinecraftForge/Installer/blob/2.0/src/main/java/net/minecraftforge/installer/DownloadUtils.java#L56
kind stirrup
Ya.. thats not what that code does.
Anyways, its a simple fix. Use the correct paths in the java args text
pearl light
Ya.. thats not what that code does.
then what does the code do? if I'm reading this properly, then libraries which are not from the Mojang maven and use that classifier-in-version trick use the path as derived from that classifier-in-version, rather than the path from path section of the library download information
kind stirrup
first it does it for libraries NOT downloaded from the mojang repo
it also does NOT use the version in path string
it uses the REAL string
Which should be in the 'path' variable in the json
Again, we've been using classified artifacts for ages, there shouldnt be anything you need to touch on this.
Maty: If you wanna give me the actual rundown of the simptom/issues let me know.
pearl light
only now do I see that small ! before url.startsWith(LIBRARIES_URL) 
damn that rabbit hole
timid kiteBOT
gohkingytp#0922 was muted. | Stop spamming the porting thread
paper hedge
So I left for a bit, anyone got a fix yet?
I guess that ender dragon problem is indicative that we should probably do a patch audit
flat zealot
I did one early on and after that looked at all patches I commited. that one sliped through because it was extremely easy to miss
I've gone to bed, so yall can figure out the server classifier stuff, I most probably won't be able to squash again so someone else can do it
paper hedge
rip
frail flare
Hmm, for some reason ChunkSerializer#write() can receive ProtoChunks which are marked as ChunkStatus.ChunkType.LEVELCHUNK, which causes them to bomb out when the chunk cap save patch tries to cast them to get the cap data: https://github.com/MinecraftForge/MinecraftForge-Private/blob/1.20-dev/patches/minecraft/net/minecraft/world/level/chunk/storage/ChunkSerializer.java.patch#L31-L39.
Link to stacktrace: https://gist.github.com/PlatinPython/3a8aee889dd2f5ebc57aa6d99aa52d1e
The patch looks identical to the one in 1.19.4 and I can't explain why a proto chunk would be marked as a full chunk
flat zealot
are you sure forge doesn't somehow make a protochunk normal
livid sequoia
Iirc, something changed in the ordering of chunk statuses for lighting, so it might be a cast that was fine in 1.19.4 can’t be assumed anymore
frail flare
The only difference I can find is that LIQUID_CARVERS (came after CARVERS) are gone, INITIALIZE_LIGHT (comes before LIGHT) is new and HEIGHTMAPS (came before FULL) is gone. Other than that, the chunk statuses appear to be the same Maybe not, the ChunkStatus#generate() method works differently now, that seems like a possible source of this
livid sequoia
Give me a sec to get to a computer
Ok, my best guess after a quick gander is that the chunk status set now set after the execution of the work rather than before
Which means we're probably doing something where the chunk was promoted to full status before it actually was
frail flare
I have a slightly different hypothesis: it looks like previously the "generation task" of each chunk status had to set the status of the proto chunk itself and ChunkStatus.FULL specifically seems to have not done that. Now, this is instead done in ChunkStatus#generate(), which means it also happens for ChunkStatus.FULL. If my understanding is correct, then this should be as simple as changing the patch I linked earlier to check instanceof level chunk
livid sequoia
We should be doing that anyways
And I agree with that, especially since the previous code block explicitly checks if it is a protochunk
frail flare
If there are no objections, I would implement that check and then I'll be off to bed as well
kind stirrup
Alright, kid dropped off. Whats going on?
frail flare
The way chunk generation states are updated changed slightly, which meant that ProtoChunks marked as full could land in the Forge patch that saves chunk caps, which does an unchecked cast and throws now
kind stirrup
Who is working on that?
frail flare
Champ and I looked into it. My suggestion is to just add an instanceof check to avoid it. I was just waiting to see whether anyone had any objections to that idea before pushing that fix
kind stirrup
Also, doesnt seem like any pushed the server installer fix.
Right now its just on the make it work phase
There is no reason we should be delaying the release
frail flare
Fair enough, then I'll push it 👍
kind stirrup
It sounds simple enough issue, a misisng type check
frail flare
Pushed. I'll be off to bed as well now.
kind stirrup
@flat zealot -^
kind stirrup
Oh didnt see he went to bed
paper hedge
the generation of win/linux_args.txt uses artifact.path which does not match what goes into the libraries folder
I'm not sure why we have the path desync besides for the URL
but it doesn't match in version.json to where it ends up in libraries folder yet the minecraft client/launcher still finds it
so maybe that's part of the bug?
kind stirrup
The launcher should be using the 'path' entry in the json
which is where we should be pushing things.
But that may not be 'Artifact.getPath' because that is based off the artifact coords
I think the issue might be a deduplication issue?
paper hedge
artifact.path is a groovy map access thing, it's the same data that goes into version.json, it's just that it's regenerated for the args.txt with different parameters
kind stirrup
Like I said we've had classified things working for a while now, But what i suspect is that one of our processors uses the api jar, and that path overrides the one in the main downloads
in voice if you wanna share/debug
kind stirrup
hehe
paper hedge
gotta get my mic setup hold
@kind stirrup I'm definitely talking so I think it's a you issue
kind stirrup
paper hedge
So 1.20 porting is completely done and so I'm gonna close this post. but it can still be searched in the future by the public 👍
pearl light
<@&1067092163520909374> 1.20.1-rc1
https://www.minecraft.net/en-us/article/minecraft-1-20-1-release-candidate-1
paper hedge
FFS
livid sequoia
Aghhghgh
frail flare
pearl light
yes
silk garden
yes
pearl light
ParchmentMC uses the in-house Axe trigger, which polls the manifest for changes and releases a notif internally
livid sequoia
Ugh, I'll worry about it in a bit
pearl light
(+ Blackstone)
silk garden
there it is
pearl light
and there's the usual notif in our channel
paper hedge
And Snowblower is broken right now, damnit
I changed the url to piston-meta and it broke downloads
pearl light
for some reason, the phrase "sucks to suck" floated through my mind
flat zealot
gahhhh
pearl light
oh hey, I forgot that Snowblower uses Axe as well
flat zealot
poor coeh
paper hedge
oof
pearl light
flat zealot
this is my chance to squash again! 
pearl light
well, back to Feather-induced madness
livid sequoia
Ok, only thing that matters was that AbstractWidget#renderWidget is now protected, which someone shouldn't be calling manually anyways
Besides from all the bug fixes
I don't think this needs a primer
flat zealot
coeh's mcpconfig action should run in 2 minutes
prisma wren
#tooling-dev message
livid sequoia
Only one change needed in Connection. It doesn't seem like there were any issues I noticed otherwise
paper hedge
Meh, since it's a bug fix release, uncommenting the forge source wasn't necessary
also that's generally done as a separate commit before you start the update
thus we have easy diffs of what Mojang changed, even if we also have snowblower. Also it means viewing the update commits don't kill your web browser
standard procedure is a track forge subprojects commit (optional) -> update mcpconfig and mc versions commit -> actually fix errors
livid sequoia
Gotcha, I will remember this for next time
golden tusk
I doubt 1.20.1 is probably going to break most mods.
It's just like 1.16.4 and 1.16.5 that are just rather minor updates fixing bugs.
livid sequoia
There’s nothing of consequence for the changes, so probably
livid sequoia
Rebased on latest Forge and cleaned up the git history so that it matches what shrimp mentioned
paper hedge
👍
livid sequoia
1.20.1 literally has no code changes
dusty wedge
I have a slightly different hypothesis: it looks like previously the "generation...
Was this the chunk status issue that was giving problems here?
pearl yarrow
That's the belief, although I don't know if anyone here confirmed it
livid sequoia
It was the leading hypothesis yeah
golden tusk
Wow 1.20.1 is already released.
winged isle
OnlyIn