reManager / vellum

[jade cedar]

that shouldn't happen...

what remarkable-os version did it generate?

[novel mesa]

3.25 or whatever

[jade cedar]

3.15?

[novel mesa]

Gimme a min to check

[jade cedar]

cool, I should be able to test that issue tomorrow for sure

[novel mesa]

Sorry yeah im on 3.25

I don't know why i thought 3.15

Is the remarkable-os package supposed to be pinned in the world file as well?

What happened was I copied the packages from my previous world file into the new one and ran vellum fix, it complained about floating-toolbar-2.0.1-r0 not satisfying the constraints, so I pinned it to an earlier version and ran vellum fix again, which installed everything, I unpinned it again and running vellum upgrade doesn't try to pull in 2.0.1 anymore

appload
btop
disable-mount-utils><Q1ATqIF7MXZkDeq0Au3EAyNpkxhGU=
fastfetch
floating-toolbar
framebuffer-spy
literm
remarkable-os
rmpp
rmstream
tailscale
vellum
vellum-bash-completion``` my world file, disable-mount-utils replaces mount-utils with scripts that don't run mount

[jade cedar]

I don’t think I’ve ever hand edited my world file 😅

[novel mesa]

It shouldn't really cause any problems afaik

I can do another reinstall tomorrow without manually editing the world file

[jade cedar]

I’ll def test out the version constraint on 3.25

[quiet folio]

vnsee attributes Khyra with one "r" too many.

[carmine frigate]

khyryra?

That's its github username

[quiet folio]

Oh. I was going by the name here. Didn't realize it was different.

I had just done a little @ with a few letters and noted the avatar and name from there. My bad. Carry on.

[quiet folio]

I feel like I saw this issue talked about already at the Vellum level, but just in case: I did upgrades through reManager that were prompted, and now:

[carmine frigate]

How can rmstream even be version dependent, it's an appload app

[quiet folio]

These are the test versions we were working with earlier too. Maybe that's part of it.

[carmine frigate]

Ah right

Interesting

[orchid valley]

That's not the case her, but appload apps can import components from the rM qmltree, if it does it can become version dependent. It could also have a backend that relies on special features of the os.

[carmine frigate]

I get that it can technically be, but the rmstream package specifically does not have a direct version requirement https://github.com/vellum-dev/vellum/blob/main/packages/rmstream/VELBUILD#L11
Maybe I should have worded that better

[jade cedar]

reManager now checks depend constraints for that filter

fbspy doesn’t support the rm1

At least the non-testing version, which is where reManager pulls metadata from

[brazen reef]

when upgrading to a new OS and then running vellum upgrade, if I want to upgrade some but not all of the packages with new versions available (say I don't want the newest version of a vellum package that doesn't have os restrictions), is the best way to do it just run vellum upgrade followed by all the packages I want to upgrade?

[jade cedar]

🤔 why not upgrade to a new version?

I don't know of a reason vellum upgrade pkg1 pgk2 wouldn't work, I believe the os mismatch handling would still trigger

[brazen reef]

mostly because of tailscale--the way it's packaged in vellum currently assumes userspace mode but I build my own kernel module which means I fudged a little bit with the post install script

not the biggest issue though, I can run one more script afterwards or keep tailscale separate from vellum

[jade cedar]

if you try that "upgrade by list" and it works, let us know!

[brazen reef]

will do (once remarkable gives me linux kernel for 3.26)

[novel mesa]

would you be open to sharing that script so i can package it into a subpackage for people that have the kernel module?

hmm i have the option of compiling vnzee with releasefast or releasesmall, releasefast gives a ~4MB binary while releasesmall gives a ~280KB binary

[jade cedar]

Ooo what’s the difference?

[novel mesa]

i truly have no idea, i dont think the performance aspect matters much so releasesmall might be more convenient for rm1 and rm2 users?

it's weird because libvncclient.a itself is 968K

yeah ill do releasesmall it runs fine for me

ah if i strip the fast binaries i get around 364KB

ill go with that

[jade cedar]

lol do upx too

[novel mesa]

nah

[carmine frigate]

@novel mesa what is

if [ -z "$force" ]; then
    verify
    initdcheck
fi

for in vnzee's velbuild?

[novel mesa]

I yoinked it from the default unpack() function from apkbuild docs

Can't remember why, might be pointless to have it in there tbh

[carmine frigate]

if you modify unpack you can add default_unpack at the start if you want it to still use the default unpacking functionality + extra

if you're fully changing it, just overwriting unpack is enough

[novel mesa]

Nah default unpack will unpack both tars and I only need 1

[carmine frigate]

which is why this isn't needed

also tar makes the folder by itself

unpack() {
    tar -C "$srcdir" -zxf "$srcdir/$CARCH.tar.gz"
}

so just this is enough

[novel mesa]

The archives dont have folders inside

[carmine frigate]

they don't have to

it's two files

just install them manually instead of cping everything

[novel mesa]

Aight

[carmine frigate]

so something like

package() {
    install -Dm755 "$srcdir"/vnzee \
        "$pkgdir"/home/root/xovi/exthome/appload/$pkgname/vnzee
    install -Dm644 "$srcdir"/external.manifest.json \
        "$pkgdir"/home/root/xovi/exthome/appload/$pkgname/external.manifest.json

    install -Dm644 "$srcdir"/LICENSE \
        "$pkgdir"/home/root/.vellum/licenses/$pkgname/LICENSE
    echo "https://github.com/0xdeb7ef/$pkgname/archive/v$pkgver.tar.gz" > \
        "$pkgdir"/home/root/.vellum/licenses/$pkgname/SOURCES
}

[novel mesa]

Yeah I'm doing it now

vnsee-qtfb has install -d "$pkgdir"/home/root/xovi/exthome/appload as the first line in package(), should I leave that in for vnzee as well?

[carmine frigate]

no, the -D flag in install makes all required folders

also

do you save anything in /home/root/xovi/exthome/appload/vnzee?

[novel mesa]

wdym by save?

[carmine frigate]

make any extra files apart from the manifest and binary

[novel mesa]

nope

[carmine frigate]

I'm actually not sure what apk's because here would be, but I'd assume that the vnzee folder will be removed by the regular uninstall anyways

[novel mesa]

probably yeah

[carmine frigate]

even tho the manifest is different

[novel mesa]

the purge is unnecessary

[carmine frigate]

@jade cedar could you publish to testing?

[novel mesa]

i do wonder if there's a way for appload apps without a backend to save configs

then i can save the user's last used hostname in the vnzee launcher

:)

[carmine frigate]

hm full screen updates feel slow

[novel mesa]

they are

[carmine frigate]

like way slower than vnsee-qtfb I mean

[novel mesa]

oh?

[carmine frigate]

let me send a video

vnzee / vnsee (fastest)

[novel mesa]

yeah that is kinda slow

i wonder why

ill have to dig into it

because i get similar speeds on vnsee and vnzee on my setup

[supple thunder]

Might be a decent idea to make a new thread for this btw given it‘s not directly reManager or vellum related?

[novel mesa]

@normal fable is it possible to write triggers for subpackages or do I have to do it the old way through files?

[normal fable]

https://github.com/Eeems/vbuild/issues/13
Not possible yet

[pallid raven]

@jade cedar say I wanna push a bug fix package, is it just an update to the vellum repo and PR that and that's it?

[jade cedar]

gotta update the checksum too

[pallid raven]

of just the qmd, yes?

[jade cedar]

yeah

[pallid raven]

alright, I've pushed an update real quick to test out the waters I guess =P

[jade cedar]

nice, you can def swap the package maintainer to you if you want

[pallid raven]

Ah, well, maybe for the next OS release

[jade cedar]

looks good, merging

[jade cedar]

just FYI, here are the xovi / qt-res-rebuilder dialogs in reManager

[brazen reef]

People really don’t read huh

[pallid raven]

if they did, IT would crumble /s

is it bad if we have them type, in words, YES I UNDERSTAND WHAT I'M DOING HERE

[supple thunder]

Didnt stop Linus from uninstalling his desktop environment, so it's not gonna make anyone else read either xD

[jade cedar]

uninstall DE
Linux sucks

[quiet folio]

I almost can’t believe he tried Pop again. Almost. At least he seems to have course corrected on at least one computer with Ubuntu or something.

[pallid raven]

Pop!_OS always be hitting at the wrong times with Linus XD

[jade cedar]

@pallid raven do you need to test the built packages for floating and gestures or you good for me to just ship em?

[pallid raven]

I did test the qmds on their own before

but maybe I should do a quick test, hmmmmm
unfortuntely, with all the rmhacks shenanigans, have probably installed so many different versions that I've lost count

[jade cedar]

omg maybe people can read

the rm2 doesn't have xxd, so my hashtab warning wasn't firing

[pallid raven]

boop, I made an oopsie, gesture had a wrong condition in it, but fixed it now

[jade cedar]

good for me to publish?

[pallid raven]

put in testing first
I'll try on my Pro as a sanity check

@jade cedar is something off on mine?
(I did have to do things manually for certain parts of qt-r-r sooo)

[jade cedar]

what version of remanager is that?

[pallid raven]

wait, good point, this is slightly old

[jade cedar]

I think there might have been a version (1.2.3 maybe?) where the filter was blocking uninstalling, which isn't ideal 😅

1.2.5 is building now with some fixes, including what I suspect tripped up "2 CS masters" guy

hard to know since they provided no data

[pallid raven]

ｒｅＭａｒｋａｂｌｅ
╺━┓┏━╸┏━┓┏━┓   ┏━┓┏━┓┏━┓┏━┓┏━╸
┏━┛┣╸ ┣┳┛┃ ┃    ┃ ┃ ┃┣┳┛┣┳┛┣╸ 
┗━╸┗━╸╹┗╸┗━┛    ╹ ┗━┛╹┗╸╹┗╸┗━╸
root@imx8mm-ferrari:~# vellum add floating-toolbar@testing
WARNING: The repository tag for world dependency 'floating-toolbar@testing' does not exist
ERROR: Not committing changes due to missing repository tags.
root@imx8mm-ferrari:~# vellum add gestures-fouzr@testing
WARNING: The repository tag for world dependency 'gestures-fouzr@testing' does not exist
ERROR: Not committing changes due to missing repository tags.
root@imx8mm-ferrari:~# vellum upgrade
No packages to upgrade.
root@imx8mm-ferrari:~#

HMMMM

wut

[jade cedar]

vellum testing status

[pallid raven]

disabled

ahh

do I uninstall the existing one?

is that what it needs

[jade cedar]

you need 3.26

right?

[pallid raven]

HMMMMMMMMM

[jade cedar]

ooo fun you're in a weird state

[pallid raven]

😭

[jade cedar]

cat .vellum/state/osver

[pallid raven]

nani the f

this SHOULD WORK?

[jade cedar]

rm .vellum/state/osver
vellum upgrade

somehow the local remarkable-os package didn't get rebuilt correctly on 3.26

[pallid raven]

root@imx8mm-ferrari:~# vellum upgrade
No packages to upgrade.
root@imx8mm-ferrari:~# vellum add floating-toolbar
Downloading 1 packages...
(1/1) Downloading floating-toolbar-1.0.7-r0
Proceeding with upgrade...
(1/1) Installing floating-toolbar (1.0.7-r0)
OK: 19.6 MiB in 15 packages
root@imx8mm-ferrari:~#

wait a min

that version ain't right

(just tried regular to see)

[jade cedar]

it's probably right for 3.24 lol

[pallid raven]

prob
but why is it not seeing that I'm on 3.26

https://tenor.com/view/cavemanspongebob-caveman-spongebob-react-whatreact-gif-20206670

[jade cedar]

what's your vellum version? vellum list -I

[pallid raven]

vellum-0.2.3-r0

so tempted to just nuke it all

random qn
no one-click uninstall?

oh nvm

found it

alright, fixed now

[pallid raven]

Works, you can publish now

[jade cedar]

I do feel like I had an upgrade state tracking bug a while back, wonder if it screwed you up

[pallid raven]

possibly

I do wonder how much my manual manipulation also screws things up

do not look into my exthome folder

[jade cedar]

reManager doesn't love when there are testing packages 😅

[pallid raven]

it's a mess with like 15 folders

[jade cedar]

luckily the "nuke and pave" approach is pretty safe here

[dull tapir]

Hi, is there a way to write local packages for vellum? I have some systemd units that I want to have installed when I call vellum reenable but that don't make sense to upstream.

[jade cedar]

certainly

you could even host your own package repo, it's just alpine apk (with some special sauce in the cli wrapper and package format, but the repo is pure apk)

[dull tapir]

oh got it, thanks

will try to set up a local repo, i'll ask again if I run into trouble

[jade cedar]

👍

the readme in the main vellum repo has info on the signing keys, building packages locally, etc. You could cheat off of the github actions if you wanted to automate publishing to your own repo

[normal fable]

https://github.com/Eeems/vbuild-action/
If you want it to be even simplier to build a single package

[dull tapir]

oh this is great, i was having issues trying to deploy my own

[dull tapir]

got it to work, but your action didn't work for me. It would always build packages as noarch, even after setting package=armv7 in the VELBUILD

I don't know if I misunderstood how vellum works, but I couldn't get it to detect noarch packages

[normal fable]

It does look like the last commit broke some tests, I can get that fixed

[dull tapir]

I have a follow up question, what's the sane way to sign the index with the local.rsa key?

i'm using a vibecoded script that reverse engineers the package format and signs it with busybox and openssl 😅

#!/bin/sh
# Sign APK packages or APKINDEX without abuild-sign
# BusyBox compatible - requires: openssl, tar, gzip, dd, wc
set -e

PRIVKEY="${1:?Usage: sign.sh <private-key> <file...>}"
KEYNAME="$(basename "$PRIVKEY" .rsa).rsa.pub"
shift

for f in "$@"; do
    FILE="$(readlink -f "$f")"
    DIR="$(dirname "$FILE")"
    SIG=".SIGN.RSA.${KEYNAME}"

    openssl dgst -sha1 -sign "$PRIVKEY" -out "${DIR}/${SIG}" "$FILE"

    cd "$DIR"
    TMPTAR="$(mktemp)"
    SIGTARGZ="$(mktemp)"
    tar cf "$TMPTAR" "$SIG"
    CUTSIZE=$(($(wc -c < "$TMPTAR") - 1024))
    dd if="$TMPTAR" bs="$CUTSIZE" count=1 2>/dev/null | gzip -9 > "$SIGTARGZ"

    TMPSIGNED="$(mktemp)"
    cat "$SIGTARGZ" "$FILE" > "$TMPSIGNED"
    chmod 644 "$TMPSIGNED"
    mv "$TMPSIGNED" "$FILE"

    rm -f "$TMPTAR" "$SIGTARGZ" "$SIG"
    echo "Signed $f"
done

I'm sure this is not the correct way to do it ahah

[jade cedar]

this is how I'm doing it: https://github.com/vellum-dev/vellum/blob/4b2701657b5d9b657d5e6202aa4ea1950290b84b/.github/workflows/build.yml#L343

[dull tapir]

i was trying to avoid dealing with the index in the CI, i'm just building a single package with my personal hooks

was hoping to curl it into the remarkable, build the index and sign it locally

[jade cedar]

you can run abuild from not CI, not from the remarkable though

[normal fable]

package=armv7 would not be how to specify a different arch, you would use arch=armv7. Did you just mispeak?

[normal fable]

same with vbuild

[dull tapir]

yes, had to double check

but I've set arch=armv7

[dull tapir]

yeah that's what I tought

welp the vibecoded script will have to do ahah

it does work somehow

[jade cedar]

abuild depends on a bunch of crap is isn't trivial to get built statically (unlike apk-tools)

[dull tapir]

cd .vellum/local-repo/armv7
curl -LO https://github.com/luishfonseca/rm-scripts/releases/latest/download/lhf-1.0.0-r0.apk
vellum --allow-untrusted index -o APKINDEX.tar.gz *.apk
curl -sL https://raw.githubusercontent.com/luishfonseca/rm-scripts/main/sign.sh | sh -s ~/.vellum/etc/apk/keys/local.rsa APKINDEX.tar.gz
vellum update
vellum install lhf

i'm installing my package like this

but yeah, thanks for the help

the both of you

[normal fable]

Odd, as vellum uses vbuild as well and hasn't had arch issues that I'm aware of. I'm checking why the tests are failing now.

[dull tapir]

I fixed it by setting the environment variable CARCH when calling vbuild

      - name: Build packages
        run: VBUILD_DRIVER=docker CARCH=armv7 vbuild -C pkg/ all

[normal fable]

Oh, that's how it's meant to be

[dull tapir]

i didn't see the CARCH in your action, is it set?

[normal fable]

No, I should fix that. But have you pulled down vbuild to look at it's --help yet?

[dull tapir]

no, i just ran it in the CI

[normal fable]

Ah, well I recommend doing so as the help output will give you more information. You can also review how vellum calls it

https://github.com/vellum-dev/vellum/blob/main/scripts/build-package.sh

[dull tapir]

yes, that was my source

[normal fable]

I'll update the action to let you specify what arch you want to build for though

[dull tapir]

My usage is so minimal that I'll leave it like this

if you get the action to generate a signed index that would be very cool btw

but probably out of scope

I tried to get an index served on github pages without success

[normal fable]

Generating a signed index would need to be it's own separate action

[dull tapir]

makes sense

[normal fable]

The action now supports setting arch

[normal fable]

@jade cedar https://github.com/rmitchellscott/reManager/issues/51
Because of this question: #1488967310679998715 message

[jade cedar]

I like it

I need to also add donation stuff and a deprecated status

Not sure if those should be first class citizens or just like _donation

[normal fable]

Hmm

well if the plan is to put them into the site as well as remanager, first class makes sense

[jade cedar]

Could start as _ then get promoted later

[normal fable]

Sure

Although maybe I get you to do the promotion work when it's ready since that is a bit of work to craft the PR for 😛

[jade cedar]

Fundamentally, reManager and the index are both gonna pull those from the json metadata so minimal changes down the road (just the metadata generator) well and every package

In my head if something is optional, it probably makes sense to leave as _-prefixed?

[normal fable]

Since _ is for local variables I wouldn't want something with that prefix to also pull double duty as a real field that is just optional

[jade cedar]

Makes sense

[normal fable]

Perhaps a metadata associative array would be in order?

[jade cedar]

Ooo

[normal fable]

It would be a little more annoying to define, but would give us flexibility, and we could figure out a way to expose the metadata so others could drive things off of it

like maybe adding metadata=(["screenshot"]="<url>") could be used for splashscreen packages and then queried by an application on device to display a preview or something

[dull tapir]

opinions on packaging https://github.com/tun2proxy/dns2socks so we can get tailscale's magic dns working on the remarkable? I'm happy to contribute it if it seems useful

my use case is keeping rmfakecloud inside my tailnet with tls enabled, currently i'm overriding the hosts file but getting magic dns to work is cleaner imo

[normal fable]

If it's not in entware already, it would be fine to open a PR to propose adding it.

[dull tapir]

i don't think it is, i'll open a PR in a bit then

[dull tapir]

https://github.com/vellum-dev/vellum/pull/131

[novel mesa]

ooh this seems useful

[dull tapir]

yeah, particularly for devices that don't support the tun module

[normal fable]

Gave you some feedback

I also triggered the build to make sure it builds

[dull tapir]

i don't really understand how the entware-rc VELBUILD works

postupgrade(){
    /home/root/.vellum/share/entware-rc/entware-rc.post-install
}

what installs this script? Is it an introspection thing and vellum detects it is needed?

[normal fable]

it's installed by apk as part of installing the package

It's the post-install script

[jade cedar]

We might have cleansed too many details from the readme for that bit

[normal fable]

It probably is worth adding to the readme how to call other install scripts from another one

Actually, it would be worth adding a way to just say postinstall and have vbuild make it work

I'll add a enhancement item to vbuild for that

[dull tapir]

i must be blind 😅

what line does that https://github.com/vellum-dev/vellum/blob/main/packages/entware-rc/VELBUILD

[normal fable]

Lines 31-39

they declare a postinstall script, which results in the entware-rc.post-install file being created automagically

and the install of it gets done automagically as well

[dull tapir]

but only gets created when required right? because i checked my system and none of my packages installed their vellum scripts there

but got it

[jade cedar]

tripletap, entware, tailscale, there are a few

Only gets created if the VELBUILD declares a postosinstall

[dull tapir]

yeah, i have those hooks in .vellum/hooks/post-os-upgrade

i was just surprised to see .vellum/share/entware-rc/entware-rc.post-install instead of the hooks dir

[normal fable]

Oh, that package might be wrong then

I was referencing the entware package when I was building it, and it's referencing scripts that are not the automatic ones

Feel free to correct entware-rc as part of your PR

The README will need an update as well: https://github.com/vellum-dev/vellum?tab=readme-ov-file#package-scripts

[dull tapir]

so should be .vellum/hooks/post-install/entware-rc?

[normal fable]

Something like that I assume, I'm not able to check at the moment

[dull tapir]

@normal fable it's getting late here, so I won't get to test it today

but thanks for the reviews

[little wren]

@jade cedar and @timid moat It would be nice when xovi and reManager use qmd files from folder named fw. Because when I tested 2 version of fw, I have to move qmd to folder disabled atc...

[jade cedar]

remanager doesn't use qmd files

[jade cedar]

tripletap has a mode that works around this

https://github.com/rmitchellscott/xovi-tripletap?tab=readme-ov-file#enable_version_switching

[little wren]

Yes, I know. reManager would move these files to the fw folder.

[jade cedar]

remanager doesn't handle files

not sure how you'd handle qmds that work on multiple versions, that sounds like a pain

[little wren]

"The system maintains separate hashtabs for each OS version"
So what is the folder structure then? Wow, you think of everything, I have to look into that.

[jade cedar]

symlinking a version-specific qt-resource-rebuilder dir

[little wren]

Ohhh, thanks - Finally I know for what is xovi/stock 🙂

OK, It looks nice.

I understand everythink, I have to after update fw remarkable 3xpress power button and then xovi-tripletap/prepare-new-version.sh
Or edit POST_START_COMMANDS=() for automation, is it true?

[jade cedar]

I think you'd want to run that command before pressing 3x after new firmware install

[little wren]

If I set PRE_START_COMMANDS and use the button even if I don't update the fw, will something bad happen?

[jade cedar]

so iirc:

• reboot into new firmware
• prepare-new-version
• rebuild hashtab
• start xovi as normal

[little wren]

Okay, I somehow got it working and it's exactly what I wanted.

But I couldn't figure out why xovi wouldn't start and it turned out to be because of appload. So shouldn't what's happening with the extension be the same on extension.d?

[jade cedar]

ah yeah appload needed a qml change between 3.25 and 3.26

I haven't built anything to screw with the extensions directory

vellum should handle the extensions downgrade/upgrade when you run vellum upgrade after booting to the new version

[orchid valley]

And unfortunately if you want all functionality you can't have it working in both

But we could make a version agnostic appload without the qmd

And then install the qmd on its own for the relevant version

[little wren]

I don't want to have all the features at the same time. I just want to test 2 versions of fw so that only what can be loaded is loaded, for version 3.25 only extensions for 3.25 and for fw 3.26 only what is compatible with 3.26.

[little wren]

  - add-finished-button
  - appload
  - bettertoc
  - disable-selection-autoscroll
  - fav-tag-button
  - fix-page-number-on-slider
  - hide-document-close
  - quicksettings-screenshot
  - random-sleep-screen
  - selection-erase
  - three-finger-swipe-to-reset-view
  - toc-from-selection
  - touch-lock

Either wait for them to be updated, or remove them with 'vellum del <package>'.
Then run 'vellum upgrade' again.```

This doesn't seem to help.

[orchid valley]

That's because you shouldn't be installing these things on a beta version

[jade cedar]

ah. people doing dev work on betas probably shouldn't use vellum at all

[orchid valley]

And I think it does help. It tells you exactly which packages (all) you shouldn't expect to work on that version

[little wren]

I don't have this in version 3.27, that's it.

So I'll explain again:
I have version 3.26 for example.
And there are add-ons.
Everything works.

Then I want to install for example 3.25 and have other add-ons there, because those from version 3.26 are not compatible with fw 3.25.

As soon as I switch back to 3.26, the add-ons from 3.26 should be loaded.

As I say @mitchell is right and xovi-tripletap works great - it's well thought out but only on xovi/exthome/qt-resource-rebuilder but it doesn't work on xovi/extensions.d

[carmine frigate]

This can be archived with a pre-start xovi script, assuming you're on v18+

qmds too, it is now xovi native and so tripletap is not required

[jade cedar]

extensions.d really didn't change often when I added the symlinking to tripletap.

[little wren]

I've read it a few times and I think I understand, I'm going to try it with 2 versions of fw that aren't beta.
OK, it'll probably work like you wrote then - yeah, it makes sense to me now. vellum upgrade

[jade cedar]

where'd vellum get 3.27.0.87 from?

[carmine frigate]

/home/root/xovi/scripts/pre-start/versioned.sh:

#!/bin/bash

os_ver=$(. /etc/os-release && echo "$IMG_VERSION")
os_ver_major_minor=${os_ver%.*.*}

rm -f /home/root/xovi/extensions.d
mkdir -p /home/root/xovi/extensions.d.versioned/$os_ver_major_minor
ln -s /home/root/xovi/extensions.d.versioned/$os_ver_major_minor /home/root/xovi/extensions.d

rm -f /home/root/xovi/exthome/qt-resource-rebuilder
mkdir -p /home/root/xovi/exthome/qt-resource-rebuilder.versioned/$os_ver_major_minor
ln -s /home/root/xovi/exthome/qt-resource-rebuilder.versioned/$os_ver_major_minor /home/root/xovi/exthome/qt-resource-rebuilder

use at your own risk, not tested

you have to store extensions and qmds like

/home/root/xovi/
  extensions.d.versioned/
    3.22/
      extension.so
    3.23/
      extension.so
    .../
  exthome/qt-resource-rebuilder.versioned/
    3.22/
      name.qmd
    3.23/
      name.qmd
    .../

and you have to remove the /home/root/xovi/extensions.d and /home/root/xovi/exthome/qt-resource-rebuilder folders first

edit: you can also symlink this script into ~/xovi/scripts/debug if you want it to work if your first xovi start is a debug start
also make sure to chmod +x the script

[little wren]

I have a beta on one section.
I'm not developing, nothing for beta - I know it's not right.
I'm going to switch to classic versions and then try again.

[dull tapir]

I got the dns2socks thing working and dig shows the expected records but xochitl can't find the rmfakecloud. It doesn't even call the proxy, I'm getting an "unspecified network error" on RequestDeviceTokenHostNameReceived

I had to go back to writing it in my hosts file

Also, my original idea that we could use this to the tailscale magic DNS was a misunderstanding of how that works, the quad100 IP is a local thing that doesn't exist on userspace mode

In my setup I wanted to reach an internal DNS server so that wasn't an issue, but definitely kills the broad appeal

[novel mesa]

yeah, you'd need the tun module for it to work

[dull tapir]

yeah bummer

[brazen reef]

I have a WIP pr to get the tun module packaged into vellum if you want to take a peek!

[dull tapir]

I'm on rm2, your changes are for the new models right?

[brazen reef]

Ah yes, unfortunately.

[novel mesa]

@normal fable can you help me rebase off your pr without messing up the commit history?

I did a rebase off pr135, but now it says i have one commit to pull (im guessing 2c0b5d3 which is the old tailscale tun support one), so the only way to push this is to force it?

[normal fable]

When you rebase, you always have to force push to replace your old copy of the branch with your new one

From the git log, it looks like you rebased properly

Github would just show more than one commit until the other PR is merged

[novel mesa]

I'll deal with it tomorrow

[novel mesa]

@normal fable did you set this up on your vbuild repo?

[normal fable]

Yeah, I use it for code review

[normal fable]

You can now call other lifecycle scripts in your VELBUILD. It'll automatically insert a helper function to call the script, including post-os-upgrade.

You can see an example here: https://github.com/vellum-dev/vellum/blob/3fb2655846cdd4a4a20f799a775db4fd19eccedd/packages/entware-rc/VELBUILD#L40-L42

[supple thunder]

Has anyone here been able to get reManager to talk to KWallet outside KDE (Hyprland in my case)? I'm not sure what I'm doing wrong but somehow I can't get it to work. It also doesn't spit out any logs, so I'm kinda just stuck right now

[normal fable]

Which xdg-portal implementation(s) do you have installed?

[supple thunder]

xdg-desktop-portal and the kde and hyprland versions

It does prompt to unlock the keychain called login but then it still errors saying it can‘t save to the system keyring

[jade cedar]

The lib I’m using is pretty limited with keyring stuff. It seems to expect a very specific configuration

But alas it’s the only lib

I’m legit shocked by the amount of Linux users of a GUI application

[supple thunder]

From what I could tell the only limitations are needing the keyring to be named „login“ and having a dbus system whatever compatible secrets service. As far as I can tell I have both but it don‘t worky

[jade cedar]

Hmm

I will never target desktop Linux again

[supple thunder]

I mean I‘m not really asking you to support this given that it seems to work on the major desktop environments. I was just hoping someone could tell me what I‘m doing wrong

[jade cedar]

It should log debug stuff to a file

[supple thunder]

o

Where would that be?

[jade cedar]

Can’t remember offhand and I’m on my phone but generating a log bundle and not uploading it will zip it up for you

[supple thunder]

Gotcha let me check that

[jade cedar]

Idk if it’ll have this specific thing logging

But can certainly add more logs if not

[supple thunder]

I typically just check terminal output for debug messages but reManager doesn‘t log anything there

[jade cedar]

Yeah I’m suppressing debug in terminal output for production builds. Or wails is, can’t remember

[supple thunder]

Fair enough but i‘d expect errors to appear there

[jade cedar]

Yeah same

[supple thunder]

Doesn't seem like there's anything in the log

Yea doesn't log anything when hitting the connect button

Err

It works now???

[jade cedar]

sus

[supple thunder]

Not sure what I changed, but I'm glad it works now haha

[brazen reef]

@jade cedar q re: createpages-paperpro and prevent-notebook-zoomout; vellum currently lists them as meant for all devices and I guess there's nothing that says you can't run these on all devices, but perhaps consider making them pro move only?

[jade cedar]

Why

rm2 / rmpp interop could be useful

[brazen reef]

well your description on github does say that especially prevent-notebook-zoomout was designed for paper pro move so that's where I was coming from

createpages-paperpro/rm2 could certainly be useful for rm2/paper pro interop, true

[jade cedar]

Currently I think I’m only enforcing device-specific when it breaks something cause it’s kind of annoying

[brazen reef]

that's fair

[novel mesa]

disregard, forgot to update vbuild

oh wait i should probably update vbuild first

alright it builds

ok the tailscale-tun subpackage is ready, im now just waiting for @brazen reef's tun module to be packaged so i can make tailscale-tun depend on it

[brazen reef]

yay in the next week or so I'll set up a proper build situation to test the tun module package and such

[jade cedar]

@novel mesa this build should have debug logs in the terminal https://files.scottlabs.io/owuserid.zip

[novel mesa]

$ flatpak run io.scottlabs.reManager
Overriding existing handler for signal 10. Set JSC_SIGNAL_FOR_GC if you want WebKit to use a different signal```

[jade cedar]

Hmm

[novel mesa]

then it just exits

the window pops up for a second then exits

[orchid valley]

@jade cedar Just tried reManager for the first time. When I had built the hashtab there was a button at the top which said, I needed to start xochitl. It was not obvious to me (until pressing that button), that I needed to go to [Maintainance] to start xovi. I suggest that the average user should not need to go to Maintainance on a first run, but that there should be a prompt at the end of rebuild_hashtable to start xovi (and a hint: you can restart in the maintaince tab at any time).

[jade cedar]

Yeah I need to rework that

I didn’t realize rebuild_hashtable stopped starting xochitl so that banner was intended to be a fallback not the primary flow

[carmine frigate]

not reading the non-existing changelogs smh

[jade cedar]

I don’t want to make starting xovi too easy or automatic

Because then the first time people think about it will be the next time their tablets reboot

[orchid valley]

Then perhaps a link on the rebuild hashtab popup: Go to maintainance and choose Xovi [start] to .... or Xovi [stock] to ....

[jade cedar]

Yep I just need to have some time to think through how I want the user flow to be

Cause I think I want to present a “oh hey you didn’t install tripletap, here’s a button to do that” too

[jade cedar]

new support UI with "report issue" and updated xovi/hashtab UI flow. Not released yet.

[pallid raven]

This might sound harsh but genuinely add an internal timer that doesn't let you click boxes immediately on a users first time interacting with this

[carmine frigate]

oh I hate that so much

[pallid raven]

Yeah but this ain't T&C

[carmine frigate]

add a cli flag so devs can disable it

actually since it's just first time

it's fine I guess

-# not that I use remanager anyways lmao

[supple thunder]

just my two cents, but I think just saying "For technical reasons, xovi don't auto start" might make some people just question it more and maybe make some bad life choices. I'd instead give said technical reason in short form, maybe something like "To avoid a situation in which your device can no longer turn on properly, UI mods [...]"

That portrays the seriousness of said technical decision better without infodumping imo

Other than that, love what you're doing with this! 10x more helpful than any other software diagnostic i've seen ngl

[jade cedar]

lol how long of a timer

[pallid raven]

Find someone who has never used this, have them read it aloud

That long

[supple thunder]

The world needs more hallway tests

[novel mesa]

make them wait 3 hours

then give them a 60 minute quiz

[carmine frigate]

"Do you solemnly swear that you will never autostart xovi?"

[pallid raven]

HARRY, DID YOU AUTOSTART XOVI, said dumbledore calmly

[normal fable]

@jade cedar I'm noticing that reManager is sending the input events, but the device is still suspending when I leave it on the pin screen. I'll see if the same happens after unlock.

[jade cedar]

Interesting, pen hover may not reset the timer when the screen is locked

Do you see the events in evtest?

[normal fable]

Yes

After unlocking I still haven't seen it lock again

Perhaps this is meant to try to protect against the power button being bumped and the device waking in your bag

[jade cedar]

I wonder if I can detect pin screen and to do either a pen tap or touch in a corner of the display. I didn’t wanna do real UI interaction but it’d be fine on the pin screen

[normal fable]

There is also always resurrecting this concept: https://github.com/Eeems-Org/sysfs_preload

[jade cedar]

I don’t really wanna modify the tablet state

[normal fable]

Fair enough

[jade cedar]

I like that with poking inputs it’s not possible for the tablet to get stuck in a prevent sleep mode once disconnected

[normal fable]

I like that too

It does look like it's something about the lockscreen

it's still unlocked and hasn't gone back to suspend mode

[brazen reef]

30 seconds?

[jade cedar]

24

[carmine frigate]

Oddly specific

[jade cedar]

autocalculated at 170 wpm

[brazen reef]

oh I assumed you just took a screen cap after a few seconds

[jade cedar]

nah I'm quick

rebuild is 17

[carmine frigate]

If you install tripletap do you still have to wait?

[jade cedar]

yes

there's a setting to disable

[carmine frigate]

Ui suggestion: make the button horizontally fill up with black as time passes

[brazen reef]

maybe give them a 5-10 second credit if they install tripletap 😛

[jade cedar]

also only the package install hashtab rebuild gets the timeout, the one from Maintenance doesn't

[jade cedar]

lol that'll just distract them from reading

[carmine frigate]

Okay another suggestion: eye tracking check if the window is active, if it isn't pause the timer /s

Like discord quests

[jade cedar]

probably need more flatpak permissions for that..

this does feel really slow to me, but

1. I've seen this hundreds of times
2. I read at like 430 wpm

[brazen reef]

170wpm does seem on the slower end

[carmine frigate]

Research done in 2012[13] measured the speed at which subjects read a text aloud, [...] 228±30 for English
While proofreading materials, people are able to read English at 200 wpm on paper, and 180 wpm on a monitor.

[jade cedar]

it def is, I did some googling for reading speeds for technical writing and that seems to be kind of around 150, but 150 felt slow so I bumped it up a bit

I mean I could just do like 10sec for them instead of math

or 5sec, idk.

[brazen reef]

who knows with these things

[jade cedar]

I am not a UI designer 😭

to me, 5s feels like enough to make people go "ah right, I should read this not just pound the next button"

[brazen reef]

but you also read in general

[jade cedar]

true

[brazen reef]

ultimately, there's only so much handholding you can do for folks, and you're already going above and beyond

[supple thunder]

Nah 5s is definitely in that „monke see timer, monke press skip once available“ region of the brain

YouTube has firmly cemented that

But if it‘s a 20s unskippable people might be bored enough to actually pay attention

[jade cedar]

or alt+tab to something else

[supple thunder]

Well what can you do

You can bring a horse to water but you can‘t force it to drink

[jade cedar]

idk a 27 sec unskippable lock-out seems user-hostile

[brazen reef]

maybe make the button read "I understand how to start mods after the tablet reboots"

I know that's not really button-length

[jade cedar]

I mean at the end of the day it’s the user’s problem

[brazen reef]

random idea: upload to the user's tablet a document "how to restart mods after reboot or upgrade"

[jade cedar]

shit I could

or just a remanager users guide in general

[jade cedar]

hmm could make a fake choice to encourage comprehension (I mean one option has a real action associated)

[brazen reef]

or you could make it like, do you want to install tripletap to easily restart mods when the tablet reboots? 1) yes (-> installs tripletap), 2) no, I'll use reManager or ssh command /home/root/xovi/start instead (->does nothing)

make tripletap the path of least resistance

similar to what you have, but as you have it people might assume they have to either stick with remanager or running ssh command

[quiet folio]

The one thing I would consider changing is bolding do not auto-start and by design. Maybe also prevent your device from booting.
-# And everything else. Bold the user. Block the user! Problem solved? In more seriousness, I'm not sure about the bolding. It's something I'd probably just mock up and see how I felt about it after. As a non-professional UI designer who has just read a ton about UI design over the years and absorbed a fraction of it.

[jade cedar]

fair point, but I don't think most remanager users are gonna be screwing with systemd

lol I just realized this doesn't solve the flow for user that already had tripletap queued for install, will have to give that some thought too

[normal fable]

https://github.com/vellum-dev/vellum/pull/146
vbuild now handles systemd units automatically (aside from template units)

[little wren]

@jade cedar Just a quick question, what does the reManager program depend on?
Don't let it die on newer drivers 🙂
I recently got a newer iMac but it's still an old 2017.

[jade cedar]

Not up to me

Wails says it supports down to 10.13, but I think someone had issues below 10.15

[little wren]

OK, maybe it will still work for a while.

[jade cedar]

Yeah I think so

10.15 supports back to some 2012 models

[supple thunder]

Wouldn't that machine also be pretty easy to update using OpenCore Legacy Patcher?

[little wren]

I don't have a good experience with this. Maybe it was because of the old hardware I had (iMac 2011).

[jade cedar]

Wails v3 (still in alpha) supports back to 10.15 as well

[median ether]

Hello ! My koreader install freezes regularly and I have to restart the whole device. It's usable, but a bit annoying. I've investigated what I could on the device.

Nothing is logged on KoReader. Everything seems fine.
But systemd logged at the time of the crashes:

then killed xochitl, koreader.sh, and both reader.lua processes with SIGABRT.

So KOReader seems to be taken down as a child of xochitl.service, not just crashing by itself.

Investigating further why xochitl might crash, I found that my xovi install is apparently corrupted and causes instability. xochitl is logging repeated xovi/QML patch errors:

• Cannot assign to non-existent property layerFloatToggle
• Cannot assign to non-existent property resetFloatBarPositionFromPrimary
• Cannot assign to non-existent property presetSaveFromPrimary
• Failed to parse file: SyntaxError: JSON.parse: Parse error (qrc:/qml/device/view/documentview/SceneViewGestures.qml:58)

gestures.qmd is also trying to parse /home/root/colorSet.json, and that file does not exist.

Is there any way to clean install xovi through vellum? I tried to do it again but I don't see a way to force the install

# vellum add xovi

OS upgraded (3.25.0.142 -> 3.25.1.1).
Run 'vellum upgrade' to sync packages with new OS version.
Run 'vellum reenable' to restore packages that modify the system partition.

Thank you very much for the help ! 🙏

[pallid raven]

The json stuff is fine

To remove it completely
Just go to settings>Accessibility and set a few colours there

I think the rest are due to

Not having floating bar

[carmine frigate]

vellum fix xovi xovi-extensions maybe?

[median ether]

Thank you @pallid raven , but it's shown as installed. Is there a way to force reinstall it?

[carmine frigate]

The second is important as xovi itself is just a .so, the xovi/* structure is provided by xovi-extensions

[median ether]

Thanks, did not know this command. I just ran it.

[pallid raven]

Ohh?
Hmmm, I don't remember seeing those errors if you do have them installed, odd

[carmine frigate]

What else do you have installed? vellum list -I

[median ether]

I've ran the vellum fix command on xovi, xovi-extensions, and floating-toolbar. Hopefully that'll get it fixed! 🤞

[median ether]

~# vellum list -I
appload-0.4.1-r1 aarch64 {appload} (GPL-3.0) [installed]
curl-8.11.0-r1 aarch64 {curl} (curl) [installed]
floating-toolbar-1.0.9-r0 aarch64 {floating-toolbar} (MIT) [installed]
gestures-fouzr-1.0.7-r1 aarch64 {gestures-fouzr} (MIT) [installed]
koreader-2025.10-r1 aarch64 {koreader} (AGPL-3.0) [installed]
mount-utils-1.0.0-r0 aarch64 {mount-utils} (MIT) [installed]
qt-resource-rebuilder-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]
remarkable-os-3.25.0.142-r0 noarch {remarkable-os} (MIT) [installed]
rm-version-switcher-1.0.0-r1 aarch64 {rm-version-switcher} (MIT) [installed]
rmpp-1.0.0-r0 noarch {rmpp} (MIT) [installed]
selection-erase-1.0.3-r1 aarch64 {selection-erase} (MIT) [installed]
selection-stuff-1.0.2-r1 aarch64 {selection-stuff} (MIT) [installed]
toolbar-icon-1.0.7-r0 aarch64 {toolbar-icon} (MIT) [installed]
tripletap-1.0.0-r1 aarch64 {tripletap} (MIT) [installed]
vellum-0.1.9-r0 aarch64 {vellum} (MIT) [installed]
vellum-bash-completion-3.0.3-r2 aarch64 {vellum-bash-completion} (GPL-2.0-only) [installed]
xovi-0.3.2-r0 aarch64 {xovi} (GPL-3.0) [installed]
xovi-extensions-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]
xovi-message-broker-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]

[carmine frigate]

Oh damn that's outdated

vellum upgrade would be a good idea, you're still using extensions v17

[median ether]

I've tried running vellum upgrade. I thought this was the way to update ^^ Damn, I should go read how to use the vellum cli 🤦

[median ether]

I did though

~# vellum upgrade
OS upgraded (3.25.0.142 -> 3.25.1.1). Checking package compatibility...

All packages have compatible versions. Preparing upgrade...
Index has 2 packages (of which 2 are new)
No packages to upgrade.

[carmine frigate]

Hm do vellum update and then upgrade

[median ether]

 [/home/root/.vellum/local-repo]
 [https://packages.vellum.delivery]
OK: 343 distinct packages available
root@imx8mm-ferrari:~# vellum upgrade
OS upgraded (3.25.0.142 -> 3.25.1.1). Checking package compatibility...

All packages have compatible versions. Preparing upgrade...
Index has 2 packages (of which 2 are new)
No packages to upgrade.```

😭

[carmine frigate]

Weird, does it even see v18 as available?
vellum list | grep xovi-extensions

[median ether]

Yes it does 🙂

root@imx8mm-ferrari:~# vellum list | grep xovi-extensions
framebuffer-spy-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
framebuffer-spy-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
qt-command-executor-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
qt-command-executor-17.0.0-r5 aarch64 {xovi-extensions} (GPL-3.0)
qt-command-executor-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
qt-resource-rebuilder-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
qt-resource-rebuilder-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]
qt-resource-rebuilder-17.0.0-r5 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: qt-resource-rebuilder-17.0.0-r4]
qt-resource-rebuilder-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: qt-resource-rebuilder-17.0.0-r4]
webserver-remote-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
webserver-remote-17.0.0-r5 aarch64 {xovi-extensions} (GPL-3.0)
webserver-remote-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
xovi-extensions-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
xovi-extensions-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]
xovi-extensions-17.0.0-r5 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: xovi-extensions-17.0.0-r4]
xovi-extensions-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: xovi-extensions-17.0.0-r4]
xovi-message-broker-16.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0)
xovi-message-broker-17.0.0-r4 aarch64 {xovi-extensions} (GPL-3.0) [installed]
xovi-message-broker-17.0.0-r5 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: xovi-message-broker-17.0.0-r4]
xovi-message-broker-18.0.0-r2 aarch64 {xovi-extensions} (GPL-3.0) [upgradable from: xovi-message-broker-17.0.0-r4]

[carmine frigate]

Maybe outdated vellum cli version then? I think versioned pacakges like xovi extnesions are handled by the wrapper and there was a bug at some point where that didn't work

Try vellum add vellum and then upgrade again

Yea I think that was added/fixed in 0.2.4, and you have 0.1.9

[carmine frigate]

Alternatively if that doesn't work re-run the bootstrap https://github.com/vellum-dev/vellum-cli#installation

[median ether]

Yeah, I'll re-run the bootstrap, the command did not work. Thank you so much for the help.

Re-running vellum upgrade fails to update the packages.

root@imx8mm-ferrari:~# vellum upgrade
OS upgraded (3.25.0.142 -> 3.25.1.1). Checking package compatibility...

All packages have compatible versions. Preparing upgrade...
Index has 2 packages (of which 2 are new)
(1/1) Upgrading remarkable-os (3.25.0.142-r0 -> 3.25.1.1-r0)
OK: 111.1 MiB in 19 packages
Failed to check for upgrades: ERROR: unable to select packages:
  remarkable-os-3.25.1.1-r0:
    breaks: appload-0.5.1-r0[remarkable-os>=3.26]
            floating-toolbar-2.0.3-r0[remarkable-os>=3.26]
            gestures-fouzr-1.0.10-r0[remarkable-os>=3.26]
            toolbar-icon-1.0.8-r0[remarkable-os>=3.26]
    satisfies: world[remarkable-os=3.25.1.1-r0]
               appload-0.5.1-r0[remarkable-os<3.27]
               floating-toolbar-2.0.3-r0[remarkable-os<3.27]
               floating-toolbar-2.0.3-r0[/bin/sh]
               gestures-fouzr-1.0.10-r0[remarkable-os<3.27]
               gestures-fouzr-1.0.10-r0[/bin/sh]
               koreader-2026.03-r0[/bin/sh]
               qt-resource-rebuilder-18.0.0-r2[/bin/sh]
               selection-erase-1.0.3-r2[remarkable-os<3.27]
               selection-erase-1.0.3-r2[remarkable-os>=3.24]
               selection-stuff-1.0.2-r2[remarkable-os<3.27]
               selection-stuff-1.0.2-r2[remarkable-os>=3.23]
               toolbar-icon-1.0.8-r0[remarkable-os<3.27]
               tripletap-1.0.0-r2[/bin/sh]
               xovi-0.3.3-r0[/bin/sh]
               xovi-extensions-18.0.0-r2[remarkable-os>=3.20]
               xovi-extensions-18.0.0-r2[/bin/sh]

Sorry again for being a bother

Should I update the rmpp to 3.26? I thought the support for 3.25 was better 🙂

[carmine frigate]

Nah you're fine, pretty sure this is a vellum issue since it's based on a package manager that only cares about the latest version of each package

[carmine frigate]

Looks like everything you have is available for 3.26, so yea go for it

3.24 for rm hacks, I think pretty much everything else already supports 3.26

[median ether]

I've updated to 3.26, updated vellum itself with the bootstrap, upgraded all packages and rebuilt the hashtable. Hopefully, this'll fix the crash.

Thank you again for the help and time you gave me. Have a nice day! 👋

[cobalt night]

yeah old version of appload is known to cause koreader system hang

[carmine frigate]

Is that present in past koreader versions too or just 2026.03+?

[cobalt night]

26.03+

[carmine frigate]

Hm but they had 25.10 (#1464668214003896435 message)

[cobalt night]

might be self updated unstable

[median ether]

It might have been, I had OTA updates. Don't know if I updated to 26.03+. I honestly don't think so, I had not updated Koreader for at least 1 or 2 months.

[cobalt night]

did you update to unstable?

those unstables in last 2mo also have issues

[median ether]

Might have been that then. I probably did

[cobalt night]

KOReader: exposing AppLoad's flaw since 2025

[timid moat]

I think I fixed that, right?

[cobalt night]

yep

it is fixed

but nevertheless it's fun to call it like that

[pallid raven]

we need to make an Appload stress test app xD

[cobalt night]

well KOReader does the job well lol

it discovered some quite nasty bugs

like subprocess quitting crashing appload and such

[timid moat]

That reMarkable allnighter is still on the backburner for me. I have quite a few things I want to fix

[jade cedar]

There was a vellum cli bug in versions older than 0.2.6 (the latest) that sometimes resulted in vellum upgrade not running correctly.

[normal fable]

Do you leave OTA updates enabled in the vellum koreader package? I would recommend against it.

[jade cedar]

I haven’t done anything with OTA I thought those defaulted to off but I don’t use it so not sure

Will have to check

[normal fable]

I believe it defaults to being available to enable, I would recommend just disabling it completely

sed -i 's/hasOTAUpdates = yes/hasOTAUpdates = no/' "$pkgdir"/opt/koreader/frontend/device/remarkable/device.lua
This is what toltec does

[jade cedar]

yeah, I'll do similar

[orchid valley]

It's a bit weird that Noa Himesaka did some work for the OTA updates to work, and that you then turn it off 😄 But I guess you want that, such that vellum can do all the updates?

[cobalt night]

for package manager managed installation it makes sense

and no I didn't work on OTA update, it just worked (after I nagged them to publish remarkable-aarch64 build)

[jade cedar]

Yeah it’s just hard to provide support when they list package versions and that reports something thats not running. Power-users can hopefully figure out how to edit the file to get on the nightly track if they want

[normal fable]

Perhaps a koreader nightly package with OTA enabled?

I guess it depends on if there are enough koreader users who would want nightly?

[normal fable]

@jade cedar https://github.com/vellum-dev/vellum/pull/167 for your review to add proper building support where the VELBUILD can specify what image to use for the build instead of running build() in the builder image that everything else runs in

[jade cedar]

Sweet!

[normal fable]

Got a sample VELBUILD in the PR for you as well

We'll see if it holds up as more packages switch over, but so far so good

Did I link the wrong thing?

Yup

too many tabs open

[pallid raven]

seems like something about the license checksum, I have messed up

[jade cedar]

probably abuild cache locally when you generated it

[pallid raven]

hmm?

cause I did a PR and it failed

[jade cedar]

I fixed the checksums and merged

also gestures was living in the 3.28 future

[pallid raven]

oh whoops

xD

[normal fable]

In case anybody missed it, this was merged. Having a VELBUILD do the package build using any container image with any toolchain is now extremely straight forward. Just specify the image= variable in the VELBUILD and the build() function will run inside that container. If you need more packages installed in an image, I recommend pushing a new image to a container registry with them preinstalled instead of installing them in the build() method. This avoids issues where a VELBUILD stops working because the base image is so out of sync with it's package repositories to install packages anymore.

[jade cedar]

I feel like we have a couple packages that could benefit from this. In the middle of 3.27 update hell at the moment though

the vellum org can also certainly host container images and build actions if needed (I think toltec did that as well right?)

[normal fable]

The netevent package I included in the PR uses the toltec toolchain image

Which is probably the image most can start using since v4.0 can build for both armv7 and aarch64

[brazen reef]

oh interesting, I wonder if I can make use of this instead of having pre-compiled tun modules hanging out on a github repo

[normal fable]

You can! It will mean that packages take longer to build, which is something to take into consideration

[brazen reef]

true, I'll poke around!

[normal fable]

https://github.com/vellum-dev/vellum/blob/main/packages/netevent/VELBUILD Here is the example VELBUILD making use of the feature

[brazen reef]

yep, looking at that now 🙂

[normal fable]

Also, if you want to see what it has to do to get it to work, take a look at the resulting APKBUILD when you do a vbuild gen. It works, but it does feel like I might be missing another better way.

[brazen reef]

will do

[normal fable]

Another change for VELBUILDS merged: https://github.com/vellum-dev/vellum/pull/175

tl;dr you can now specify readmeurl, donateurl, and status. These aren't reflected on the site or remanager yet, but that is coming. The PR took care of updating all the existing VELBUILDS that made sense with their readmeurl and some other housekeeping.

[jade cedar]

Gonna start working on reManager 1.5 with those fields and initial Pure support.

@pallid raven make sure to get your donate link added to yo packages

[late sequoia]

Why does this break? I am trying to install restream:

Errors occurred:
Proxy download failed: failed to simulate install: ERROR: unable to select packages:
xovi-message-broker-18.0.0-r4:
breaks: quicksettings-screenshot-1.0.2-r4[xovi-message-broker<=17]
satisfies: world[xovi-message-broker]
rm-shot-1.0.2-r0[xovi-message-broker]
rmstream-0.1.4-r1[xovi-message-broker<19.0.0]
rmstream-0.1.4-r1[xovi-message-broker>=18.0.0]

[jade cedar]

Uninstall quicksettings-screenshot probably

It needs an older version of xovi-message-broker but restream needs a newer one

This channel is also for development of reManager / Vellum, not support.

[jade cedar]

vellum.delivery updated with status, donate links, and readme parsing (and Paper Pure support). reManager 1.5.0 will likely release tomorrow with the same. I’ve added donateurls for the ones I knew offhand.

[cobalt night]

KOReader marked unmaintained? huh?

also it shouldn't be marked rMPPure compatible yet

[late sequoia]

oh my bad. Sorry

[jade cedar]

Oof something screwed up lol

I’ll take a look and get that fixed shortly

[jade cedar]

metadata-generation fixed (marking pkgs unmaintained when they weren't), rmppure exclusion going in for a couple more packages:

• koreader
• vnsee
• fbspy

[normal fable]

I wonder if it would make sense to have vellum also manage entware packages if it's installed, so users can just vellum add nnn

At the very least having reManager have a section for listing/installing entware packages would probably be helpful

[jade cedar]

it would also be nice to be able to add entware packages as depends, I have not spent much time thinking through that though

lol the scope creep in reManager also becoming an entware UI

[normal fable]

Should be possible with vbuild just adding in some pre-install and post-remove scripts

[jade cedar]

idk that reManager's UI/UX is optimized for an additional 3000 packages 😵

[normal fable]

I've spent too much time in my life retrofitting bad list views to support larger datasets than designed 🙁

One client had a to do list meant for a couple tens of items filled with like 20k items. Just a fundamental misuse of the list, but they were not willing to change their business.

@jade cedar #general message
Does the javascript bundle parse the contents of <noscript> at all?

[jade cedar]

Nope completely separate

[normal fable]

In that case, perhaps it would be better to just include a redirect in the noscript to a page that has no JS to avoid forcing everybody to download it all but not use it every time?

[jade cedar]

Probably, this is what I shit out in like 30 minutes lol

[normal fable]

Fair enough.

I tend to use redirects like this:

<noscript>
  This page requires JavaScript, for a simplier view that does not, please go <a href="<url here>">here</a>
  <meta http-equiv="refresh" content="0; url=<url here>"/>
</noscript>

[jade cedar]

It’s like 50KB on top of 1.3 MB of js

[normal fable]

but the js can be cached, the page does not appear to be

In fact, it's explicitely set to not allow caching the index

cache-control: public, max-age=0, must-revalidate

[jade cedar]

Hmm

Could be my overly aggressive rule to prevent cloudflare from caching the actual packages

[normal fable]

Probably

[jade cedar]

I’ll keep looking at it

[jade cedar]

been working on a mock-up of a nextgen Mods view. Thoughts?

[plucky hearth]

I like it

only thing bothering me is sideways scrolling, but if it#s only for the overview it should be fine

[jade cedar]

yeah the other pages are normal

[plucky hearth]

I like

[jade cedar]

I need to add a list view back in for the non-overview pages

[jade cedar]

just a lil heads up: if anyone is running reManager 1.5.2, there was a bug in the metadata processing that resulted in the xovi and qt-res-rebuilder commands not showing up in maintenance. I pulled the release (only out for a couple hrs), about 20 people grabbed it from GH while it was up, idk about flatpak. Fixed in 1.5.3.

[normal fable]

1.5.3 is perpetually telling me that mods are installed but not running, even though they are

and clicking the x doesn't close it

[jade cedar]

Hmm I’ll look.

[jade cedar]

I'm detecting that via the presence of /etc/systemd/system/xochitl.service.d/00-xovi.conf, that exists for you right?

[normal fable]

It does not, and after clicking the "Start UI with Mods" button it still does not

[jade cedar]

that seems odd, and xovi is actually running?

[normal fable]

Yes, appload etc are in the sidebar

I'm on 3.25 right now, I'll update and see if it's still a problem on later versions

Also, only testing on the rM with the broken screen right now

Hmm, the upgrade detected prompt should also show on the maintenance page, as the normal update etc buttons don't do it, and there was no big warning at the top that it needed to be done

@jade cedar the file exists on the latest OS, so your detection method doesn't work for 3.25

[jade cedar]

OS upgrade only shows up on the mods page

[normal fable]

Right, which as part of the user workflow where there is a big warning at the top, you miss that as you click the button to go to maintenance to fix the other things without realizing you need to click the OS upgrade first

[jade cedar]

ah, is the confusion that there's an Upgrade button on the maintenance tab?

I'm probably moving that to the Mods tab with the UI update

[normal fable]

It's that you don't even see the upgrade os button at all as your eyes are drawn to the bit warning at the top, so you click the button to take you the maintenance and then start working your way through clicking the buttons to fix the issues it lists

and then you can't get the UI to start with mods and the upgrade button doesn't update any files

and only when you go to the mods page do you get prompted to do a vellum os upgrade

[jade cedar]

I wonder what the ideal flow for that is

[normal fable]

The quickest I can think of would be to have another big warning at the top about the OS upgrade that people deal with first

A more ideal flow is probably a wizard that walks you through things in the correct order, or better yet a "You just updated, let us finish up things" dialog that just applies all the normal things like the vellum os upgrade, timezone, etc

[jade cedar]

yeah wizard would be nice

I think I didn't do that at first cause I wasn't checking as many states, but now I have a state check for almost everything so why not

[jade cedar]

NOTE: If broken hardware causes any errors or defects, such errors or defects will not be covered by reManager

[normal fable]

@jade cedar can I poke you to review the PRs I opened early this morning?

I've started to work through the toltec packages to see what makes sense to port

[jade cedar]

I’m camping tonight / tomorrow but can take a look once I’m back! Feel free to ping me whenevs!

[normal fable]

Enjoy!

[normal fable]

I've gotten to this same state in 3.27.1.0 where /etc/systemd/system/xochitl.service.d/00-xovi.conf doesn't exist, but xovi for sure has loaded. My guess is that when you install something that touches /etc, mount-utils discards any overlay changes

[tranquil elbow]

hi all, force-koreader-landscape works for 3.27.1 however when i uninstall to remove the error i cannot re-install. How can i go about force installing it? the obvious --help commands listed still complain about os version

or indeed a better way for the move to stay in sync with the device rotation

[jade cedar]

You can’t force it. That qmd doesn’t have a version published for 3.27

@orchid valley

[orchid valley]

I accidentially renamed the file for 3.27, but it's still there, and has been there from day 1. Renamed back.

[jade cedar]

ah okay, I missed that!

are any of the other new ones ready for packaging?

[orchid valley]

I think they will be later today (I do have some bugfixes from today, which are not pushed yet), I can give you a list when they are. For addPalettes.qmd I've chosen to load qml files from /home/root/colorPalettesnot sure how easy that would be to add as a package (my guess is "very").

I was actually hoping to add LOAD ingatellentSettings.qmdin all mods which required that - as that is supported in the current codespace for qmldiff, but I don't think it's in the releases for qt-resource-rebuilder yet, so I think I'll skip that for now.

[jade cedar]

should colorPalettes get stored in exthome to keep ~/ clean?

but yeah, adding some qml files as a package is ez

[orchid valley]

Well, isn't exthomemostly for xovi extensions (myExt.so)? But I do agree, that it could get a better placement. Currently all my external files are in ~/: ~/sleepScreens,~/stickers,~/colorPalettes

[jade cedar]

you're doing LOADs for that one?

[orchid valley]

for which one? currently I'm not doing any loads in puplished qmd's

[jade cedar]

colorPalettes

[orchid valley]

No, but I am using a Loader to load plaintext qml files.

[jade cedar]

ah

[orchid valley]

Here are the mods, I consider ready for packaging
addAspectLockToSelection.qmd
addCompletedButton.qmd (renamed from addFinishedButton, sorry, Completed sounds better)
addPalettes.qmd (still using ~/, though - and I suggest to package the folder colorPalettesand the contents in the package - if you think I should put palettes in exthomeinstead, let me know, and I'll change that)
closeFolderOnCtrlUp.qmd
enableLandscapeSleep.qmd
fixNavigateUpShortcut.qmd
fourFingerSwipeForLastDoc.qmd (requires ingatellentSettings.qmd)
fourFingerSwipeLeftForCompleteAndOpenNextDoc.qmd
shortcutForSplitView.qmd (requires splitView.qmd)
showToolOnChange.qmd
splitView.qmd

Feel free to filter out anything, you feel is to tailored towards my own needs.

I still consider addColorSelector.qmd and addRotationButton.qmd, addEditsToSelection.qmd to be alpha releases (well for addEditsToSelection it's mostly the required xovi extension, I'm not too satisfied with).

[jade cedar]

ooo split view is ready? I assume that still conflicts with floating toolbar?

[orchid valley]

I've no idea, but I think it should mostly work - it works with the stock toolbar with only a proxy for the position manager. If you don't want to have the floating toolbar changing tools in both views, you probably need to add a check for activeFocusor something.

There are still some minor bugs for keyboard navigation, but they are actually from the stock software, I just haven't fixed them (and then they become more prominent in split view)

[jade cedar]

fourFingerSwipeLeftForCompleteAndOpenNextDoc.qmd what a name

oh so fixApploadRotation and forceKoreaderLandscape are the same? goddammit lol

[jade cedar]

@orchid valley are ^ those 2 actually the same? If so which should I remove?

[normal fable]

@jade cedar when vellum is running an action, it should also keep suspend from happening

I just had it suspend while I was installing debian-chroot

[jade cedar]

oof

a wakelock should be enough right? It doesn't actually need to prevent xochitl sleep I think

[normal fable]

Not sure, xochitl uses sysfs to sleep, not sure if that respects wakelocks

[jade cedar]

at least, a wakelock will keep ssh running when xochitl enters sleep

were you over wifi?

[normal fable]

usb

[jade cedar]

hmm, usb should set the charger wakelock

[normal fable]

It took a while before it went to sleep, but it did

[jade cedar]

my move sets this:

root@imx93-chiappa:~# cat /sys/power/wake_lock
udev.charger

[normal fable]

Same

[jade cedar]

well bummer that'd have been too easy

[normal fable]

I've replicated installing things that use mount-utils to remount /etc for changes breaks remanager's running with mods check

[jade cedar]

is that with restarting xovi after?

[normal fable]

no

[jade cedar]

ah okay

[normal fable]

Remounting /etc removes the temporary file you are checking to see if it's running

[jade cedar]

makes sense then, it's tmpfs is gone but it's still running

I'll have to see if I can detect that more cleverly

[normal fable]

it might make sense to have a simple mod that exposes a socket you can ping to see if the mods are running

named socket in kernel space probably

[supple sparrow]

Are you trying to detect if xovi is running?

[jade cedar]

ye

[supple sparrow]

Maybe a post-start script creating a file in a known location with the pid of xochitl inside?

So you check if it exists and if the process with that pid is running

[jade cedar]

tr '\0' '\n' < /proc/$(pidof xochitl)/environ | grep -q LD_PRELOAD.*xovi works

[normal fable]

That does work

although what's nice about a named socket that has to respond is that you can also check to see if it's frozen for some reason

[normal fable]

@jade cedar https://github.com/vellum-dev/vellum/pull/202 any concerns? Just did a full add/del/add/purge cycle to make sure it all worked

[jade cedar]

lgtm

[orchid valley]

Yes, they are the same. I'll try my best to keep calling it forceKoreaderLandscape, as that's what it does (one can of course adapt it to other appload apps to fix their rotation, but I'm not planning on doing that - it doesn't make sense for apps, where you can't rotate the displayed text within the app).

[normal fable]

@supple sparrow https://github.com/vellum-dev/vellum/pull/204 for you to review as the maintainer of the package.

[orchid valley]

Why is it the old version with huge marging (aspect ratio = original) that's packed? There is a newer codebase where the gdc fits nicely on the move screen (and displays on the other devices as a 16:9 window with small margins left and right https://github.com/timower/rM2-stuff/tree/tilem-full. Unless something changed since I tried it, that is.

[normal fable]

It's using the latest release, not a non-released branch

Someone should really poke timower to release the branch, I can update to use a commit from the branch instead I guess

That said, the branch is 96 commits behind the main branch

@valid sphinx could you make a new release that adds the full option to tilem?

[orchid valley]

Okay, that makes sense

[supple sparrow]

I'm not sure what the whole git sequence is for
if the only thing we're getting with LFS is the icon, why not just add
https://media.githubusercontent.com/media/timower/rM2-stuff/refs/heads/tilem-full/apps/tilem/draft/tilem.png (with the right release / commit)
to the source and use the .tar.gz?
also as ingatellent already mentioned, I think using the main branch loses a large part of the functionality, since it can't really be used as a floating window

[orchid valley]

@jade cedar Sorry, splitView.qmd needed an update (which I've just done): Found out that the version of appload I tested against, was outdated. I've now added two lines which makes it run with appload - but unfortunately not without appload. I'm not really sure how to make a version that will run both with and without, but I'm assuming that most users will be using appload anyway.

[jade cedar]

I haven't packaged any of the new ones yet

[normal fable]

For the icon, it's so we use as much from the upstream source as possible, and as the upstream source changes and potentially adds more to LFS, it'll just work.

timower reacted to my message asking for a new release, so perhaps we'll get the functionality added to the main branch in a release soon 🙂

[supple sparrow]

@normal fable btw, you're free to either take any of my packages or set them as unmaintained, I'm not planning on contributing to vellum anytime soon

[orchid valley]

I should probably figure out how to that myself, but that's not going to happen anytime soon (it's not even on my to do list with rM related stuff).

[valid sphinx]

v0.1.4 is out!

[jade cedar]

I like that this solves our silly suffix issue too

[normal fable]

Working on integrating now

[normal fable]

And live

also in case anybody missed it, you can now install yaft with vellum. We are aware that typing doubles up, I'm fairly certain it's due to bugs with appload's input forwarding

[jade cedar]

ah bummer. vellum's sleep inhibit in 0.3.0 hangs then times out on my rm2. Pushing 0.3.1 now to revert till I can find a better method and test it more throughly

[orchid valley]

@jade cedar Just tried to update an rM2 with wifi connection from 3.26 to 3.27 using reManager (according to the ui v1.3.0), got this error:

(null) 0 file I/O error (null):0 - file I/O error
[ERROR] : SWUPDATE failed [0] ERROR : Error reading configuration file /tmp/swupdate.cfg
Error parsing configuration file: cannot read, exiting.```
Updated to v1.5.3 got the exact same error.
```ls /sys/devices/platform/
Fixed MDIO bus.0  consumers         imx-cpufreq-dt    power             reg-dummy         regulatory.0      suppliers         uevent```
Since a factory reset it has never updated except for me running `swupdate-from-image-file` to update from 3.25 to 3.26.

It looks like the error is the ouput from swupdate-from-image-file, so it might not be reManager's fault...

[jade cedar]

That def looks like it’s coming from swupdate-from-image-file, but I can take a look and make sure it’s getting served the right file

[orchid valley]

The file looks right. I am wondering why the file system doesn't have /sys/devices/platform/lpgpr/root_part...

[jade cedar]

that's a paper pro thing isn't it?

[orchid valley]

Aah... So the file name looks right

but might not be...

Downloading my own image file and running swupdate-from-image-file remarkable-production-image-3.27.1.0-rm2-public.swu works

[jade cedar]

I'll test on my rm2 tomorrow my time

[supple sparrow]

That's normal

Happens on all devices because rm uses the same script for all of them

It's a check to stop downgrades from 3.22+ to 3.21- on the paper pros

[orchid valley]

(@jade cedar Upgrading packages with no wifi is also not completely foolproof ERROR: add-finished-button-1.0.3-r1: DNS: transient error (try again later) ERROR: add-tags-to-context-menu-1.0.2-r1: DNS: transient error (try again later))

[jade cedar]

Probably not. It’s a huge pain in the ass

[orchid valley]

I know 😄 It's just that reManager installed these mods without wifi on the tablet, so I think it should also be able to upgrade them (or at least tell me to connect to the internet in some way). Well. I just needed a 3.27 hashtab, so I'm good for now. I'll be downgrading again, so I can probably test a fix, if you make one for the upgrade process.

[jade cedar]

It should be able to

What vellum version are you on?

I feel like there was a bug in a previous one around upgrades (maybe as recent as 0.2.6)

[orchid valley]

vellum v0.2.5

There doesn't seem to be any button for updating the vellum version (yes, I'm stubbornly trying to see how far I get with the reManager ui)

[jade cedar]

Ah, should update with all the other packages

should

[orchid valley]

Okay. vellum upgrade had no effect, but vellum vellum upgrade did tell me there was an update (0.3.1-r0) which it then couldn't download...
But manually downloading the most recent vellum apk installing with vellum add vellum.apk, worked. It could then not upgrade the packages directly (transient DNS error), but using reManager it worked. So I guess that bug is fixed.
How velleum knows there's an update without wifi goes beyond me...

[normal fable]

Did you do a vellum update yet?

[orchid valley]

No. I was sort of expecting reManager to handle everything, so I'm just trying to report the behaviour as an unknowing user.

[normal fable]

Fair enough, if vellum update was run and it was connected, then it would know about an update. Otherwise if reManager did something similar for transferring the updated repo files it would know

[orchid valley]

@jade cedar The upgrade from 3.26 to 3.27 worked this time. However, I was never prompted to run "Upgrade packages" during the upgrade process (I was directed to Maintaince, to rebuild hashtab, "reenable" vellum, but I was never directed to the Mods tab to upgrade the packages before pressing Start UI with mods).

[orchid valley]

Okay, yeah, that's probably it - that makes sense.

[jade cedar]

yeah I need to build a new flow for this

[jade cedar]

How vellum knows there's an update with wifi...
reManager places an APKINDEX cache

[normal fable]

If I add something like rm-pokemon to the .bashrc without an interactive guard reManager fails to load. It might be worth handling this sort of senario and not loading bashrc for when you execute commands

[jade cedar]

idk that I can skip it

[normal fable]

--norc --noprofile don't work?

[jade cedar]

In the inner shell but not the outer

rmdevlogin hardcodes —login

[normal fable]

Hmm, well that's not a thing on the rM1 or 2

[orchid valley]

Just used reManager to add appload to a freshly updated rM1. After intalling I was directed to the Maintainance tab, but the qt-resesource-rebuilder button was not there, untill I restarted reManager. The header did tell me to rebuild the hashtab, but not button was visible.

[jade cedar]

You weren’t prompted to rebuild the hashtab in the package install flow?

[orchid valley]

I didn't install anything else but appload, so I don't think I was. I was then asked to go to the maintainance tab, where I could change the timezone and turn off auto-updates, but nothing about qt-resource-rebuilder there.

I was asked to install triple tap, but I refused that

So I'm not 100% sure, I didn't skip a dialog prompting me to rebuild the hashtab, but I'm completely sure, that there was no such possibility in the application until I restarted.

[supple sparrow]

@normal fable re: tilem
https://github.com/vellum-dev/vellum/blob/main/packages/tilem/VELBUILD#L55
in my fork that was used before the native build I changed the rom download url from ipfs.io to archive.org, becuase the former is blocked by some ISPs
maybe change it to archive.org here as well? the file is the same, the source is just more trusted in general

also https instead of http

[normal fable]

I think it's probably worth having that change upstreamed instead of mucking with it in the package build

[supple sparrow]

I don't think https can be upstreamed tho cause wget

[normal fable]

reMarkable: ~/ /usr/bin/wget --no-check-certificate https://web.archive.org/web/20231217022138if_/https://tiroms.weebly.com/uploads/1/1/0/5/110560031/ti84plus.rom
Connecting to web.archive.org (207.241.237.3:443)
saving to 'ti84plus.rom'
ti84plus.rom         100% |**************************************************************************************************************************************************************************************| 1024k  0:00:00 ETA
'ti84plus.rom' saved

Seems to work on 3.27 on the rM2

[jade cedar]

wget pulls https fine lately. It just doesn't check certs

[supple sparrow]

Not on below 3.20 for some sites

[jade cedar]

"lately"

[supple sparrow]

Github fails completly there

[jade cedar]

a year ago is ancient to me

[normal fable]

In that case, upstream probably should have branching logic there that depends on what OS version it's run on

Or drop support for older, up to timower though

[supple sparrow]

I don't love that tbh, maybe just http + a hardcoded checksum (deleting the rom on fail)?

[supple sparrow]

Maybe ping them?

[normal fable]

Feel free to do so

[valid sphinx]

It's just this you want changed?

-      "http://ipfs.io/ipfs/QmSNmqjQ1Ao4jff9pXbzv98ebuCzKf2ZuyEpSGL31D8z44";
+      "https://web.archive.org/web/20231217022138if_/https://tiroms.weebly.com/"
+      "uploads/1/1/0/5/110560031/ti84plus.rom";

[normal fable]

I believe that is all they are asking for, although you'd need to add --no-check-certificate to wget's args if you haven't already for it to work on newer OS versions.

[jade cedar]

🍿 scanly has been submitted

[normal fable]

I saw that

[jade cedar]

and the dude's already DM'd me twice on reddit asking how long approvals take..

I don't like that it touches ~/.config/remarkable/xochitl.conf
I don't know enough about appload's display handling to know if what they're doing with qtfb is wrong but it feels weird and nonstandard to me

[normal fable]

If it's still using a temp file to determine what waveform to use, they are not using qtfb correctly as that's part of the API to ask for specific waveforms

Looks like they still are using it

As for touching xochitl.conf just for OrientationLocked should be fine

[jade cedar]

does xochitl read that without a reboot?

I don't understand why they have a run.sh that's setting envs external.manifest supports..

[normal fable]

It's the backing store for QSettings, so yes it should

[normal fable]

Same

[jade cedar]

yeah, xochitl doesn't read that value during normal runtime lol

is there an example of a qtquick app running via appload?

[normal fable]

I'm not sure. Since Qt apps all build with the old version of libqsgepaper and qt5, it's not possible to port them directly yet

[normal fable]

Editing ~/.config/remarkable/xochitl.conf without a xochitl restart doesn't do anything
@jade cedar not completely true, I do know that historically I could cause a crash when editing the wifi state. It all depends on when a https://doc.qt.io/qt-6/qsettings.html#sync call is made to save and load changes.