Stable merge for week 29 of 2021

[matteodelabre]

Important notices:

• When upgrading, users should run opkg update && opkg upgrade twice because of the changes introduced by this merge. (Implement architecture separation #310)
• Toltec no longer supports OS releases earlier than 2.6. You can check your current OS version in the settings panel. If you’re still running on 2.5 or earlier, please upgrade to 2.6 or newer after upgrading your Toltec install. (Update packages for release 2.6 and onwards #366)

Less important notices:

• To re-enable Toltec after a system upgrade, run toltecctl reenable instead of the previous entware-reenable script that has been removed with this update.
• You can now easily uninstall Toltec by running toltec uninstall.
• This release marks the end of our work towards transitioning to a new repository layout. This means that stable will be updated more frequently from now on. If you switched to the testing branch because stable has been outdated for a few months, and if you prefer to run stable releases, you can go back to the stable branch by running toltecctl switch-branch stable.

Layout change: The repository is now split into architecture subfolders. The rmall subfolder contains packages that can run both on reMarkable 1 and reMarkable 2 without modification. The rm1 and rm2 subfolders contain packages that are specific to each of the two devices. This change requires updating the Opkg configuration to point to the appropriate architectures. The migration is automated by placing an upgrade to the toltec-bootstrap package in the root folder of the repo that contains a migration script.

New packages:

• bash-completion - 2.11-3 (Add bash-completion and toltec-completion packages #277, Fix typo in bash-completion #375)
• display - 1.0.1-7 (replaces the server part of the old rm2fb package) (Implement architecture separation #310)
• fuse - 1.0.0-3 ([gocryptfs] Add fuse kernel module #331)
• gocryptfs - 2.0-beta2 ([gocryptfs] Add gocryptfs package (build without openssl) #318)
• innernet-client - 1.3.1-2 ([innernet] wrapper for wireguard #350, Update innernet to v1.3.0 #358, [innernet] Update to v1.3.1 #361)
• netsurf - 0.4.0-2 ([netsurf] Add netsurf package #339)
• puzzles - 0.2.2-3 ([puzzles][0.2.2] add simon tatham's portable puzzle package #326)
• remarkable-stylus - 0.0.3-1 (Create remarkable-stylus package #383)
• rm2fb-client - 1.0.1-7 (replaces the client part of the old rm2fb package) (Implement architecture separation #310)
• tilem - 0.0.7-2 ([TilEm][0.0.6] New Package #352)
• toltec-completion - 0.2.0-1 (Add bash-completion and toltec-completion packages #277)
• yaft - 0.0.4-4 ([Yaft][0.0.4] - New Package #261, [Yaft] Set build type to release #353)
• zoneinfo-utils - 2021a-2 (Create zoneinfo-utils package #182)

Updated packages:

• appmarkable - 0.0.0-11
• bufshot - 0.1.0-5
• calculator - 0.0.0-15
• chessmarkable - 0.8.0-1 (Update chessmarkable to 0.7.0-1 #362, Update chessmarkable to 0.7.1-1 #373, Update chessmarkable to 0.8.0-1 #392)
• ddvk-hacks - 23.02-1 (Update ddvk-hacks to latest #385)
• draft - 0.2.0-20
• evtest - 1.34-3
• fbink fbdepth fbink-doom - 1.23.2-1 (Update fbink to 1.23.2 #387)
• fingerterm - 1.3.5-14
• genie - 0.1.5-3
• harmony - 0.1.3-3
• iago - 0.1.0-4
• keywriter - 0.1.0-3
• koreader - 2021.07-1 (KOReader 2021.03 #330, Update KOReader to 2021.04 #348, Update KOReader to v2021.05 #357, Update KOReader to v2021.06 #378, Update KOReader to v2021.07 #389)
• lamp - 0.1.0-4
• libdlib libdlib-dev - 19.21-2
• libvncserver libvncclient libvncserver-dev - 0.9.13-2
• mines - 0.1.2-4
• nao - 0.1.3-3
• oxide erode fret rot tarnish decay corrupt anxiety - 2.2-1 (Update to Oxide v2.2 #382)
• plato - 0.9.17-2 ([plato] Update to 0.9.17-1 #344)
• quickjs - 2020.11.08-2
• recrossable - 0.0.0-7
• remarkable-splash - 1.0-5
• remux - 0.1.9-4
• restream - 1.1-2
• retris - 0.6.3-3
• rmservewacominput - 0.3.0-2
• simple - 0.1.4-3
• templatectl - 0.1.3-3
• toltec-bootstrap - 0.1.0-1 (includes toltecctl) (Manage Opkg configuration with toltecctl #379)
• vnsee - 0.4.0-2 (Update VNSee to 0.4.0 and add launcher files #355)
• wikipedia - 0.1.0-3
• wireguard - 1.0.20210219-2
• xochitl - 0.0.0-7 (Create remarkable-stylus package #383, Update to Oxide v2.2 #382, Fix xochtil chroot launching #386)
• zshelf - 0.3.1-3

Website: The source files and scripts used for generating the website have been transferred to a separate repository (https://github.com/toltec-dev/web). A website rebuild is triggered automatically from this repository’s workflow when the stable branch is updated, so that the displayed checksum of the bootstrap script is always accurate.

[matteodelabre]

The build artifacts are hosted at https://toltec-dev.org/stable-next to facilitate testing. Starting from an existing stable install, edit /opt/etc/opkg.conf and replace https://toltec-dev.org/stable with https://toltec-dev.org/stable-next to simulate the roll-out of this merge.

[matteodelabre]

Testing on rM2, version 2.8:

• puzzles: When starting a game or when the program tries to autosave, I can see a line saying that it can’t read or write in /opt/etc/puzzles/save. The game state is therefore not restored as it should when exiting and re-entering a game. This can be fixed by creating the /opt/etc/puzzles/save folder (although it might make more sense to store this in ~/.cache?) (puzzles: Fix missing save dir #394).
• netsurf: Is missing the rm2fb ELF patch, so running it directly from a SSH session without prefixing it with rm2fb-client doesn’t work (netsurf: Improve rM2 compatibility #395).
• tilem: Is downloading the TI ROM over HTTP, it might be better to use HTTPS here. Also it would be advisable to store the ROM and the state saves outside of the /home/root folder so that it doesn’t get too crowded. ([tilem] Download ROM over https timower/rM2-stuff#7)
• fbink, fbdepth, & fbink-doom: Are missing the rm2fb ELF patch too, it seems. At least, I couldn’t make them work without prefixing them with rm2fb-client (fixed in Update fbink to 1.23.2 #387).
• harmony: I get a SIGABRT when I open the Load menu if it contains previously saved images (harmony: segmentation fault in the “Load” dialog rmkit-dev/rmkit#136).
• remux: I can’t start Xochitl (blank screen) (remux: Xochitl sometimes doesn’t start rmkit-dev/rmkit#135).

Not tested:

• innernet (couldn’t make it work, but this is probably because I don’t know enough about innernet itself)
• remarkable-stylus (I don’t have a Lamy pen)

[Eeems]

Some of these issues I would consider blockers. Have you opened issues for any of them yet?

[Eeems]

@raisjn any ideas on what's happening with remux?

[matteodelabre]

Some of these issues I would consider blockers. Have you opened issues for any of them yet?

Not yet, but I plan to do it very soon. I’ll add links in my previous comment when they are open.

[matteodelabre]

any ideas on what's happening with remux?

This only happens when running remux directly from the command line instead of as a systemd service. So it’s not a major bug if we expect that most users will use it as a service.

[Eeems]

any ideas on what's happening with remux?

This only happens when running remux directly from the command line instead of as a systemd service. So it’s not a major bug if we expect that most users will use it as a service.

Ah, so I'm guessing it's just missing environment variables. If it's working fine as a service, which is the expected use-case, I no longer consider this a blocker.

[raisjn]

This only happens when running remux directly from the command line instead of as a systemd service. So it’s not a major bug if we expect that most users will use it as a service.

what is the bug? is it something to do with rm2fb or something in remux's env variable setup? i'd prefer not to ship bug if we can avoid it (because down the line will still have to field this request of 'broken remux' but have even less details)

[Eeems]

what is the bug? is it something to do with rm2fb or something in remux's env variable setup? i'd prefer not to ship bug if we can avoid it (because down the line will still have to field this request of 'broken remux' but have even less details)

From #391 (comment)

• remux: I can’t start Xochitl (blank screen) [EDIT: Only when run directly from the command line instead of as a service.]

Since this is likely a pre-existing issue, I'm not as concerned about shipping it, especially since it's better than shipping nothing and having stable remain "broken".

[matteodelabre]

The remux bug proved difficult to consistently reproduce, but in the end it can happen regardless of whether remux is running as a service or not. I opened a bug report with more details: rmkit-dev/rmkit#135

[Eeems]

So what is remaining to be tested? I think the primary focus should be for testing the upgrade of apps from existing stable installs to make sure they still work.

[matteodelabre]

I’ve only tested on rM2 so far, we would also need to confirm that everything works on rM1.

[matteodelabre]

Also, shall we take this opportunity to change the rM2 support badge to “supported” on the README? I think rM2 support is pretty much on par with rM1 support now.

[Eeems]

Also, shall we take this opportunity to change the rM2 support badge to “supported” on the README? I think rM2 support is pretty much on par with rM1 support now.

I agree

[Eeems]

Alright, has anyone done an upgrade from stable with this on a rM1 yet?

[matteodelabre]

Not yet. Just updated https://toltec-dev.org/stable-next with a new build of this branch, by the way.

[matteodelabre]

Completed testing on rM1. The only issue I found is the following: after installing gocryptfs and fuse, I needed to manually run modprobe fuse before gocryptfs path/ would work. Wasn’t the depmod -a command included in the configure script of the fuse package meant to automate this?

[Eeems]

Completed testing on rM1. The only issue I found is the following: after installing gocryptfs and fuse, I needed to manually run modprobe fuse before gocryptfs path/ would work. Wasn’t the depmod -a command included in the configure script of the fuse package meant to automate this?

It is meant to automate the module loading. It might not work at updating a module that's already been loaded though.

[matteodelabre]

This happens when first installing the module, without it being loaded first.

[Eeems]

This happens when first installing the module, without it being loaded first.

Very odd. I do seem to remember that it was working last I tested.

[matteodelabre]

I think we need to run systemctl restart kmod-static-nodes && systemctl restart systemd-tmpfiles-setup-dev after depmod -a in the configure() script so that /dev/fuse gets created. Once /dev/fuse is present, the kernel module will get automatically loaded whenever a program tries to access that device (see #405).

[Eeems]

It probably isn't needed for the stable merge for now. Anything else outstanding? Or should I approve the PR?

[matteodelabre]

Since it’s a rather simple fix that I’ve confirmed to work I’d rather add it to this merge, if you don’t mind. This may save us some user reports that the module doesn’t work.

[matteodelabre]

This being done, I think we’re ready for the merge.

[Eeems]

⛵it!