Turris OS 5 is coming

Is it true that Turris OS 5 already coming soon?

It has been in works for while in HDB branch.

What do you consider soon, e.g. one week, one month, …?

Weeks, only three unresolved issues in GitLab and moving others to 5.0.1 or 5.1.

There are few more things than the milestones, latter I would trust being indicative development targets but not necessary implying specific time frames for stable releases.


OpenWrt inadvertently introduced a major bug at the last moment (two days) prior the release of 19.07 [1] (which TOS5.x is currently based on) that caused some grievance not only for TOS packages and resolving patches were only committed in OpenWrt yesterday.
It took considerable time/effort by the TOS developer to debug and report the matter.
Now that that the patches are available the TOS development has to discover whether all pertinent issues are cured or whether there is still some residue outfall or even new issues being introduced.


Then there is the TOS workflow [2] to be taken into consideration

, which current exhibits

Annotation%202020-01-25%20210450

and thus implies the TOS5.0 release to be based on an upcoming OpenWrt 20.3 release.
At this stage/date 20.3 only indicates OpenWrt’s potential intention to fork (in March 2020) a stable branch from their development branch but the actual release date might be farther in the future; as seen for 19.07 it took nearly six months between forking and actual release [3]


Last but not least the matter of automatic migration from TOS3.x [4] that may or may not be a deciding factor for the release date of TOS5.x.


[1] https://bugs.openwrt.org/index.php?do=details&task_id=2723
[2] https://gitlab.labs.nic.cz/turris/turris-build/blob/master/WORKFLOW.adoc
[3] OpenWRT started to fork 19.07 from Master
[4] https://gitlab.labs.nic.cz/turris/turris-os-packages/issues/344

Turris OS 4 is based on OpenWrt 18.06 and Turris OS 5 can be based on OpenWrt 19.07.

The workflow however states otherwise and the developers likely will have a sound reasoning for it (which is beyond my knowledge).

That workflow is generic workflow. It is not suppose to tell you anything about versions. You are looking to much in to that graph.

The point is that branches are there and workflow describes how versions flow trough those branches.

At the moment HBK is 4.0 branch, HBL 5.0 and for some time HBD is also 5.0. In future HBD will be transformed to 6.0. Further in future HBK is going to became 5.0 while HBL 5.1. Every major release of TOS is based on new major release of OpenWrt. We do not plan at the moment to skip any OWrt release.

4 Likes

TOS 4.0: 18.06 with kernel 4.14
TOS 5.0: 19.07 with kernel 4.14
TOS 6.0: 20.xx with kernel 4.19

Can only read what is available in the public domain and draw a conclusion from it.


That is not entirely clear at the current stage since OpenWrt

is currently conducting a poll of whether forking 20.3 with kernel 4.19 or skipping that date and fork later in 2020 (July ?) with kernel 5.4 instead.
  1. Selecting kernel version for the next release

The 19.07 release was delayed by a few months, so this has affected the
subsequent release as well, impacting previously selected LTS kernel 4.19 as
another LTS kernel 5.4 has been released in the meantime. So we’re now facing
a tough choice, which kernel should we use in the next release.

Pros for kernel 5.4 (released 2019-11-24, projected EOL 12/2021):

  • staying closer to upstream 5.x version, where all the development happens
  • lower maintenance effort, decreasing the amount of backported patches
  • longer upstream support period
  • we would have more time to rebase/test ramips/layerscape targets from 4.14 to 5.4

Pros for kernel 4.19 (released 2018-10-22, projected EOL 12/2020):

  • it’s possible to have 20.03 release
  • we’re not partially throwing away many months of work and testing
  • it has been tested for some time already on most platforms (ramips and
    layerscape missing)

Question: What is your preference for kernel version in the next release?

1.1 [ ] I’m in favor of having a release soon (20.03) with 4.19 kernel
1.2 [ ] I’m in favor of skipping kernel 4.19 and focusing on the 5.4 kernel
for the next release

Current results [6] are nearly a draw


[5] http://lists.infradead.org/pipermail/openwrt-adm/2020-January/001253.html
[6] https://cryptpad.fr/code/#/2/code/view/PvVBSUc8jbD8gQxcOmfrwudZCA+0eD7Xbp53FrG+5uk/embed/present/

It is reasonable, same as 4.19 backport for 19.07.

kernel 4.19 will not become available in any future 19.07.x point (maintenance) release, it will be:

  • either a new fork (release) 20.03 with kernel 4.19, or
  • a new fork 20.x (07 perhaps) with kernel 5.4

I know. It was only my personal wish.

Yeah, I would wish kernel 5.4 being available now considering its LTS release date being 2019-11-24 and already two months old. By the time it makes its way into OpenWrt -> TOS it will be almost one year gone by.

But with better support for SFP!

What is your trouble with SFP then?

Huawei SmartAX MA5671A ONT SFP for GPON networks not working.

Not sure whether you perhaps read [7] and the conclusion?

And then there is [8] which however would need to be uplifted to OpenWrt | TOS stable branches.


[7] Supported SFP modules
[8] https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=1c16b574c4f77a30a0268acee30be96ae0dc5948

To me a bigger argument for 5.4 seems that 4.19 is announced to EOL at the end of 2020 already.

1 Like

Well, they will be cutting it close with a branch off in July, unless they bork it up the same way as 19.07.

They could also stay with 4.14 with EoL by Jan, 2024. :smirk:

There is much missing in 4.19 … does not make sense to deploy it now, but then 5.4 is not even introduced yet in the OpenWrt Master branch.

Debian 10 using 4.19. It will be prolonged same as 3.16 for Debian 8 with EOL in June 2020.

https://wiki.debian.org/LTS