1.1.0rc2 available for testing!

1.1.0rc2 available for testing!

[Ludovic Courtès]

[-- Attachment #1: Type: text/plain, Size: 2316 bytes --]

Hello Guix!

I’ve run “make release” from the new ‘version-1.1.0’ branch and uploaded
the result:

  https://web.fdn.fr/~lcourtes/software/guix/1.1.0rc2

Alternately (and preferably), you can get it over IPFS.

  ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD
  ipfs get …

If you’ve never used IPFS before, first run:

  guix install go-ipfs
  ipfs init

All the files are signed with my key, also used to sign this message.

Please test things as much as you can, be it the binary installation
tarball on a “foreign” distro or the ISO image.  Once you’ve downloaded,
just follow the same instructions as before (if you want to use
‘guix-install.sh’, you’ll have to adjust it to use the right tarball):

  https://guix.gnu.org/download/

If you encounter an issue, please report it to bug-guix@gnu.org,
explaining what you did and what you observed.

Unlike rc1, this includes the i686 installation image and the VM image.
List of changes since RC1:

  $ git log --oneline v1.1.0rc1..v1.1.0rc2
  935ef44b39 (tag: v1.1.0rc2) nls: Update.
  27a1e4fd10 build: Cap build parallelism on i686.
  f4ec5e1a98 gnu: gdm: Don't offer GNOME when it's not in the system profile.
  557e6820a7 installer: Load uvesafb kernel module.
  c635610bc6 tests: Run guided installation tests from an ISO image.
  66b53c2c72 installer: tests: Don't install to a CD/DVD.
  74f01059cb vm: 'system-disk-image' honors #:substitutable? for ISO9660 images.
  abc33218f1 gnu: sdl-pango: Shorten file name of patch.
  17edf577d5 gnu: akonadi: Shorten file name of patch.
  18ffd2f968 gnu: binutils: Shorten file names of MinGW patches.
  0bd7a6bad9 gnu: installer: Fix issue with "Esperanto" locale.
  2281a77a3c gnu: xfce: Fix typo.
  325707fabe services: gnome: Fix ‘gnome’ field name.
  e830c2a8de vm: Transparently compress iso9660 images.
  08b14ab20e maint: 'release' targets runs ./bootstrap for new version string.
  eb0352e5b0 Add missing services/linux.scm to gnu/local.mk.
  f9adf35595 Update NEWS.
  4fc2132741 gnu: qemu: Skip tests on i686-linux.

If everything goes well, we can publish the release on Tuesday, 14th,
with few or no changes beyond translation updates on the ‘version-1.1.0’
branch.

Thanks in advance!  :-)

Ludo’.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Re: 1.1.0rc2 available for testing!

[Danny Milosavljevic]

[-- Attachment #1: Type: text/plain, Size: 237 bytes --]

Hi Ludo,

On Sun, 12 Apr 2020 11:19:52 +0200
Ludovic Courtès <ludo@gnu.org> wrote:

>   ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD

ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD
Error: merkledag: not found

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: 1.1.0rc2 available for testing!

[Danny Milosavljevic]

[-- Attachment #1: Type: text/plain, Size: 397 bytes --]

On Sun, 12 Apr 2020 11:35:52 +0200
Danny Milosavljevic <dannym@scratchpost.org> wrote:

> Hi Ludo,
> 
> On Sun, 12 Apr 2020 11:19:52 +0200
> Ludovic Courtès <ludo@gnu.org> wrote:
> 
> >   ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD  
> 
> ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD
> Error: merkledag: not found

Whoops, sorry, does work after "ipfs daemon".

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: 1.1.0rc2 available for testing!

[pelzflorian (Florian Pelz)]

On Sun, Apr 12, 2020 at 11:19:52AM +0200, Ludovic Courtès wrote:
> Alternately (and preferably), you can get it over IPFS.
> 
>   ipfs ls QmeLSpj3bBV5vwieVqYkbarTxmbiiP2iAaaN6rzBPLqqRD
>   ipfs get …
> 
> If you’ve never used IPFS before, first run:
> 
>   guix install go-ipfs
>   ipfs init
> 

I also needed “ipfs daemon &” as you wrote for rc1.

Regards,
Florian

Re: 1.1.0rc2 available for testing!

[pelzflorian (Florian Pelz)]

On Sun, Apr 12, 2020 at 11:19:52AM +0200, Ludovic Courtès wrote:
> If everything goes well, we can publish the release on Tuesday, 14th,
> with few or no changes beyond translation updates on the ‘version-1.1.0’
> branch.

I believe <https://bugs.gnu.org/40590> should block the release (?).

Regards,
Florian

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hi,

"pelzflorian (Florian Pelz)" <pelzflorian@pelzflorian.de> skribis:

> On Sun, Apr 12, 2020 at 11:19:52AM +0200, Ludovic Courtès wrote:
>> If everything goes well, we can publish the release on Tuesday, 14th,
>> with few or no changes beyond translation updates on the ‘version-1.1.0’
>> branch.
>
> I believe <https://bugs.gnu.org/40590> should block the release (?).

This bug is not about rc2, is it?  WDYT Mathieu?

Ludo’.

Re: 1.1.0rc2 available for testing!

[Mathieu Othacehe]

Hey,

> This bug is not about rc2, is it?  WDYT Mathieu?

No, I think its a Guile-Parted bug, I forget about, that causes
installer failures when the partition size is > 1 TiB. It has presumably
always been around.

I did a Guile-Parted 0.0.3 release and updated the package definition on
master. However, I don't have the hardware to test the fix. So maybe we
shouldn't block the 1.1.0 for this and wait for more feedback.

Mathieu

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hi,

Mathieu Othacehe <m.othacehe@gmail.com> skribis:

>> This bug is not about rc2, is it?  WDYT Mathieu?
>
> No, I think its a Guile-Parted bug, I forget about, that causes
> installer failures when the partition size is > 1 TiB. It has presumably
> always been around.

OK.

> I did a Guile-Parted 0.0.3 release and updated the package definition on
> master. However, I don't have the hardware to test the fix. So maybe we
> shouldn't block the 1.1.0 for this and wait for more feedback.

Sounds good.

Also, if we switched to 0.0.3, we’d have to do another RC or two to make
sure it’s alright.  So if it’s a corner case, I’m all for keeping the
fix for the next release.

Thanks,
Ludo’.

Re: 1.1.0rc2 available for testing!

[Maxim Cournoyer]

Hello Mathieu,

Mathieu Othacehe <m.othacehe@gmail.com> writes:

> Hey,
>
>> This bug is not about rc2, is it?  WDYT Mathieu?
>
> No, I think its a Guile-Parted bug, I forget about, that causes
> installer failures when the partition size is > 1 TiB. It has presumably
> always been around.
>
> I did a Guile-Parted 0.0.3 release and updated the package definition on
> master. However, I don't have the hardware to test the fix. So maybe we
> shouldn't block the 1.1.0 for this and wait for more feedback.

Apologies if I'm confusing things, but is this bug (about partition size
greater than 1 TiB the same as what I reported here:
https://gitlab.com/mothacehe/guile-parted/-/issues/1) ?

Maxim

Re: 1.1.0rc2 available for testing!

[Mathieu Othacehe]

Hello Maxim,

> Apologies if I'm confusing things, but is this bug (about partition size
> greater than 1 TiB the same as what I reported here:
> https://gitlab.com/mothacehe/guile-parted/-/issues/1) ?

I completely missed your report sorry about that. I need to activate
Gitlab notifications I guess :).

This is an unrelated issue that should be fixed with
eee637cab918b0f889edcdd7884970af6d8719cb. NVME disks were unsupported in
Parted 3.2, that I used when writing Guile-Parted.

Ludo, if we are on the way to a 1.1.0-rc3, maybe I could update
Guile-Parted to 0.0.4 and we could hope for installer support on NVME
disks and > 1 TiB disks. WDYT?

Thanks,

Mathieu

Re: 1.1.0rc2 available for testing!

[Maxim Cournoyer]

Mathieu Othacehe <m.othacehe@gmail.com> writes:

> Hello Maxim,
>
>> Apologies if I'm confusing things, but is this bug (about partition size
>> greater than 1 TiB the same as what I reported here:
>> https://gitlab.com/mothacehe/guile-parted/-/issues/1) ?
>
> I completely missed your report sorry about that. I need to activate
> Gitlab notifications I guess :).
>
> This is an unrelated issue that should be fixed with
> eee637cab918b0f889edcdd7884970af6d8719cb. NVME disks were unsupported in
> Parted 3.2, that I used when writing Guile-Parted.

Very nice!  Thanks for the follow-up and for your work on the installer.

Maxim

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hi Mathieu,

Mathieu Othacehe <m.othacehe@gmail.com> skribis:

>> Apologies if I'm confusing things, but is this bug (about partition size
>> greater than 1 TiB the same as what I reported here:
>> https://gitlab.com/mothacehe/guile-parted/-/issues/1) ?
>
> I completely missed your report sorry about that. I need to activate
> Gitlab notifications I guess :).
>
> This is an unrelated issue that should be fixed with
> eee637cab918b0f889edcdd7884970af6d8719cb. NVME disks were unsupported in
> Parted 3.2, that I used when writing Guile-Parted.
>
> Ludo, if we are on the way to a 1.1.0-rc3, maybe I could update
> Guile-Parted to 0.0.4 and we could hope for installer support on NVME
> disks and > 1 TiB disks. WDYT?

Sorry, I missed this message yesterday so I was stuck on the one where
you said Guile-Parted 0.0.3 and > 1 TiB disks can wait.  My apologies!

(It’s really hard for me to track all these issues and remain in sync!)

I guess 1.1.1 will fix these problems and others.

Thanks,
Ludo’.

Re: 1.1.0rc2 available for testing!

[Mathieu Othacehe]

Hey Ludo,

> Sorry, I missed this message yesterday so I was stuck on the one where
> you said Guile-Parted 0.0.3 and > 1 TiB disks can wait.  My apologies!
>
> (It’s really hard for me to track all these issues and remain in sync!)
>
> I guess 1.1.1 will fix these problems and others.

Well I didn't insist because those patches are hardly tested, and there
are some remaining issues discovered by Keyhenge anyway. So its fine
this way :) Congrats for the new release!

Hopefully the installer will be more stable for the next release, and
we will avoid the last minute patching.

Thanks,

Mathieu

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hey Mathieu,

Mathieu Othacehe <m.othacehe@gmail.com> skribis:

>> Sorry, I missed this message yesterday so I was stuck on the one where
>> you said Guile-Parted 0.0.3 and > 1 TiB disks can wait.  My apologies!
>>
>> (It’s really hard for me to track all these issues and remain in sync!)
>>
>> I guess 1.1.1 will fix these problems and others.
>
> Well I didn't insist because those patches are hardly tested, and there
> are some remaining issues discovered by Keyhenge anyway. So its fine
> this way :) Congrats for the new release!

Alright!

> Hopefully the installer will be more stable for the next release, and
> we will avoid the last minute patching.

Yup!  Anyhow, thanks for all the debugging and bug-fixing!

Ludo’.

Re: 1.1.0rc2 available for testing!

[Robert Smith]

Hi Ludo,

>   https://web.fdn.fr/~lcourtes/software/guix/1.1.0rc2

I ran into the same issue that Jesse encountered in December, the
installer hangs when it tries to connect to wifi, and I can't escape by
pressing Alt-F2/F3/F12. In the end I just have to hold down the
power button. I've tested this with different wifi networks and it's
always the same. I tried using the console to connect to wifi, and
everything worked smoothly. The only difference from the manual was
that I had to use `rfkill unblock wifi` before I could bring the
interface up. I didn't have this issue when I first installed guix, so
I downloaded the 1.0.1 installer and as I thought the issue wasn't
there, and the wifi connection went smoothly.

I'm not sure how I can get any diagnostic information from my system, if
you have ideas I'd be happy to try them out. I tested this on my Librem
15 v4; I remember from Guix Days that a few other people have Librem
laptops, I'm curious if they can use the installer to connect to a wifi
network.

-Robert

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hello,

"Robert Smith" <robertsmith@posteo.net> skribis:

>>   https://web.fdn.fr/~lcourtes/software/guix/1.1.0rc2
>
> I ran into the same issue that Jesse encountered in December, the
> installer hangs when it tries to connect to wifi, and I can't escape by
> pressing Alt-F2/F3/F12.

When you’re on tty1 (the installer), you have to hit Ctrl-Alt-F12
etc. (note “Ctrl”).  Is it what you tried?

> In the end I just have to hold down the power button. I've tested this
> with different wifi networks and it's always the same. I tried using
> the console to connect to wifi, and everything worked smoothly. The
> only difference from the manual was that I had to use `rfkill unblock
> wifi` before I could bring the interface up. I didn't have this issue
> when I first installed guix, so I downloaded the 1.0.1 installer and
> as I thought the issue wasn't there, and the wifi connection went
> smoothly.

OK.  It could have something to do with the issue about Connman not
starting that Florian reported.

What does /var/log/messages show?  (In particular, messages from
shepherd, connmand, and dbus.)

> I'm not sure how I can get any diagnostic information from my system, if
> you have ideas I'd be happy to try them out. I tested this on my Librem
> 15 v4; I remember from Guix Days that a few other people have Librem
> laptops, I'm curious if they can use the installer to connect to a wifi
> network.

Anyone else with a Librem laptop can give it a try?  You don’t need to
run the full installation, it’s enough to go past the network setup
phase, which is early on.

Thanks for your report,
Ludo’.

Re: 1.1.0rc2 available for testing!

[Robert Smith]

[-- Attachment #1: Type: text/plain, Size: 655 bytes --]




> When you’re on tty1 (the installer), you have to hit Ctrl-Alt-F12
> etc. (note “Ctrl”).  Is it what you tried?

...no, I forgot about that! Well at least that makes my job easier. I
attached the relevant section of the messages as an attachment, but
the key line seems to be:

Apr 13 17:18:52 localhost connman-vpnd[446]: Can't load 
/gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so: 
/gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so: 
undefined symbol: __vpn_ipconfig_foreach

> Thanks for your report,
> Ludo’.

No problem!
-Robert

[-- Attachment #2: guix-install-messages.txt --]
[-- Type: text/plain, Size: 2223 bytes --]

Apr 13 17:18:34 localhost installer[402]: running step 'network' 
Apr 13 17:18:34 localhost installer[402]: running step 'select-technology' 
Apr 13 17:18:34 localhost installer[402]: running step 'power-technology' 
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {update} flags 36931 <UP,RUNNING>
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 address 18:CF:5E:F4:06:75 mtu 1500
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 operstate 0 <UNKNOWN>
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {update} flags 36867 <UP>
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 address 18:CF:5E:F4:06:75 mtu 1500
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 operstate 2 <DOWN>
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 address 18:CF:5E:F4:06:75 mtu 1500
Apr 13 17:18:34 localhost connmand[363]: wlp1s0 {newlink} index 2 operstate 2 <DOWN>
Apr 13 17:18:34 localhost wpa_supplicant[362]: WPS: Converting push_button to virtual_push_button for WPS 2.0 compliance
Apr 13 17:18:35 localhost installer[402]: running step 'connect-service' 
Apr 13 17:18:52 localhost dbus-daemon[357]: [system] Activating service name='net.connman.vpn' requested by ':1.17' (uid=0 pid=444 comm="connmanctl ") (using servicehelper)
Apr 13 17:18:52 localhost dbus-daemon[357]: [system] Successfully activated service 'net.connman.vpn'
Apr 13 17:18:52 localhost connman-vpnd[446]: Connection Manager VPN daemon version 1.38
Apr 13 17:18:52 localhost connman-vpnd[446]: Can't load /gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so: /gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so: undefined symbol: __vpn_ipconfig_foreach
Apr 13 17:18:52 localhost connman-vpnd[446]: lo {newlink} index 1 operstate 0 <UNKNOWN>
Apr 13 17:18:52 localhost connman-vpnd[446]: wlp1s0 {create} index 2 type 1 <ETHER>
Apr 13 17:18:52 localhost connman-vpnd[446]: wlp1s0 {update} flags 36867 <UP>
Apr 13 17:18:52 localhost connman-vpnd[446]: wlp1s0 {newlink} index 2 address 18:CF:5E:F4:06:75 mtu 1500
Apr 13 17:18:52 localhost connman-vpnd[446]: wlp1s0 {newlink} index 2 operstate 2 <DOWN>

Re: 1.1.0rc2 available for testing!

[Ludovic Courtès]

Hi Robert,

Robert Smith <robertsmith@posteo.net> skribis:

> Apr 13 17:18:52 localhost connman-vpnd[446]: Can't load
> /gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so:
> /gnu/store/vclzrvbxac8ipc8g1ncq5gjjj8gdvxw3-connman-1.38/lib/connman/plugins-vpn/wireguard.so:
> undefined symbol: __vpn_ipconfig_foreach

I would think this one is harmless, but /var/log/messages shows that the
installer doesn’t go beyond the ‘connect-service’ step, is that correct?

Thanks,
Ludo’.

Re: 1.1.0rc2 available for testing!

[Robert Smith]

> I would think this one is harmless, but /var/log/messages shows that the
> installer doesn’t go beyond the ‘connect-service’ step, is that correct?

That's correct, it freezes on the "Connecting to the network" message.
There's no output in /var/log/messages after the lines that I attached.

Re: 1.1.0rc2 available for testing!

[Tobias Geerinckx-Rice]

[-- Attachment #1: Type: text/plain, Size: 1098 bytes --]

Robert,

Robert Smith 写道：
> I ran into the same issue that Jesse encountered in December, 
> the
> installer hangs when it tries to connect to wifi, and I can't 
> escape by
> pressing Alt-F2/F3/F12.

You need to press C-M-Fn to escape the kmscon, M- on its own won't 
do.

> In the end I just have to hold down the
> power button. I've tested this with different wifi networks and 
> it's
> always the same. I tried using the console to connect to wifi, 
> and
> everything worked smoothly. The only difference from the manual 
> was
> that I had to use `rfkill unblock wifi` before I could bring the
> interface up. I didn't have this issue when I first installed 
> guix, so
> I downloaded the 1.0.1 installer and as I thought the issue 
> wasn't
> there, and the wifi connection went smoothly.

I've run into this as well; there's an ugly patch of mine floating 
around that explicitly calls ‘rfkill’ on the interface even though 
Connman claims to do so automatically.

I'll dig up the 2-card machine that can reproduce this.

Kind regards,

T G-R

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

bug#38735: 1.1.0rc2 available for testing!

[Tobias Geerinckx-Rice via Bug reports for GNU Guix]

[-- Attachment #1: Type: text/plain, Size: 550 bytes --]

Tobias Geerinckx-Rice 写道：
> Robert Smith 写道：
>> I ran into the same issue that Jesse encountered in December, 
>> the
>> installer hangs when it tries to connect to wifi, and I can't 
>> escape
>> by
>> pressing Alt-F2/F3/F12.
>
> You need to press C-M-Fn to escape the kmscon, M- on its own 
> won't do.

Ugh.  A new (=old) subthread magically appeared after I sent this. 
Ignore this stale advice.  :-)

In better (=worse) news, I can also reproduce the hang, not just 
the rfkill issue…

To be continued,

T G-R

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Compressed ISO image got bigger

[Ludovic Courtès]

Hi,

Ludovic Courtès <ludo@gnu.org> skribis:

>   https://web.fdn.fr/~lcourtes/software/guix/1.1.0rc2

>   e830c2a8de vm: Transparently compress iso9660 images.

Here’s a comparison of the size of the ISO images for rc1 and rc2:

--8<---------------cut here---------------start------------->8---
$ du -h release-1.0.1.13450-01d5f2/guix-system-install-1.0.1.13450-01d5f2.x86_64-linux.iso.xz release-1.1.0rc2/guix-system-install-1.1.0rc2.x86_64-linux.iso.xz
260M    release-1.0.1.13450-01d5f2/guix-system-install-1.0.1.13450-01d5f2.x86_64-linux.iso.xz
405M    release-1.1.0rc2/guix-system-install-1.1.0rc2.x86_64-linux.iso.xz
$ xz -d < release-1.0.1.13450-01d5f2/guix-system-install-1.0.1.13450-01d5f2.x86_64-linux.iso.xz | wc -c
1524205568
$ xz -d < release-1.1.0rc2/guix-system-install-1.1.0rc2.x86_64-linux.iso.xz | wc -c
589223936
--8<---------------cut here---------------end--------------->8---

As you can see:

  • The xz-compressed ISO of rc2 is 1.6 times bigger than that of rc1.

  • The uncompressed ISO of rc2 is 2.6 times smaller than that of rc1.

It’s not surprising that xz has a hard time compressing an
already-compressed image, but I find it surprising that the end result
is almost twice as big.

“make release” uses the default xz compression ratio, -6.  I tried
different levels and different compression tools (though I don’t want to
change from xz now because this would likely break documentation and
links all around) and it seems we can’t really do any better:

--8<---------------cut here---------------start------------->8---
560M    /tmp/t.iso
415M    /tmp/t.iso.gz
406M    /tmp/t.iso.lz-6
412M    /tmp/t.iso.xz-0
407M    /tmp/t.iso.xz-3
404M    /tmp/t.iso.xz-8
--8<---------------cut here---------------end--------------->8---

Thoughts?  Ideas?

Ludo’.

Re: Compressed ISO image got bigger

[Jonathan Brielmaier]

On 13.04.20 16:42, Ludovic Courtès wrote:
>   • The uncompressed ISO of rc2 is 2.6 times smaller than that of rc1.

That is nice.

> Thoughts?  Ideas?

In my understanding the size of the xz compressed iso is only relevant
for downloading. So a matter of bandwith and your internet connection.

The extracted iso will be written on USB drives/DVDs/CDs, so their the
size matters and its good to see it shrinked down! So I don't see a real
problem here...

Re: Compressed ISO image got bigger

[Ludovic Courtès]

Hi,

Jonathan Brielmaier <jonathan.brielmaier@web.de> skribis:

> On 13.04.20 16:42, Ludovic Courtès wrote:
>>   • The uncompressed ISO of rc2 is 2.6 times smaller than that of rc1.
>
> That is nice.
>
>> Thoughts?  Ideas?
>
> In my understanding the size of the xz compressed iso is only relevant
> for downloading. So a matter of bandwith and your internet connection.
>
> The extracted iso will be written on USB drives/DVDs/CDs, so their the
> size matters and its good to see it shrinked down! So I don't see a real
> problem here...

I think the download size also matters, though to a lesser extent.

It’d be ideal if we could have our cake and eat it too, but at any rate
the ISO size improvement is welcome!

Ludo’.

Re: Compressed ISO image got bigger

[Tobias Geerinckx-Rice]

[-- Attachment #1: Type: text/plain, Size: 722 bytes --]

Jonathan, Guix,

There's no way around this: zisofs is an old (2004-ish?) standard 
and it's limited to zlib.  Calling XZ on the result as we do now 
is still worthwhile but not efficient.

Ludovic Courtès 写道：
> I think the download size also matters, though to a lesser 
> extent.

This is also my opinion.

> It’d be ideal if we could have our cake and eat it too, but at 
> any rate
> the ISO size improvement is welcome!

Cake will be served once I (or someone else) get squashfs boot 
working reliably.  We'll no longer *have* an ‘uncompressed’ ISO at 
that point, we'll boot directly from XZ.

But that first requires support for (real) unencrypted /boot.

Kind regards,

T G-R

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

Re: Compressed ISO image got bigger

[Ludovic Courtès]

Hi!

Tobias Geerinckx-Rice <me@tobias.gr> skribis:

>> It’d be ideal if we could have our cake and eat it too, but at any
>> rate
>> the ISO size improvement is welcome!
>
> Cake will be served once I (or someone else) get squashfs boot working
> reliably.  We'll no longer *have* an ‘uncompressed’ ISO at that point,
> we'll boot directly from XZ.

Nice!

> But that first requires support for (real) unencrypted /boot.

The ISO is not encrypted though.  Or you mean “uncompressed”?

Ludo’.