Feature #15632

Consider distributing an alternative and minimal ISO image

Added by Rover11 2018-06-03 07:20:35 . Updated 2018-09-13 17:24:37 .

Status:
Rejected
Priority:
Normal
Assignee:
Category:
Target version:
Start date:
2018-06-03
Due date:
2019-01-01
% Done:

0%

Feature Branch:
Type of work:
Discuss
Blueprint:

Starter:
Affected tool:
Deliverable for:

Description

Revisiting Bug Bug #15536, I guess the idea shouldn’t be to copy the ISO on a CD, the idea is valuable if to download, transfer, and use Tails as fast as possible.
Living in a country with strong censorship (interestingly Tails website is not censored yet, the Tor website is), and the need for open communication, a lighter version of Tails will serve to its aim. where internet speed is very low, and wide-spread Tails usage is eminent. Building Tails is impossible for the public, and downloading Tails ISO for every release from behind a regional firewall is very time consuming; A Tails-light build for every Tails release will promote its usage greatly. Considering that persistent packages may be on the way, a Tails-core (Tails-light) with only core functionality will be a great option.
I assume it is possible to build an automated script to exclude regional language support and some unnecessary (from POV of a whistle-blower in a politically oppressed situation) packages for every release. If one of the purposes of Tails is to help people to communicate freely, these features seem limitations rather than features for those in need of fast access:
Regional support: Removing all languages from Firefox, Thunderbird…(Tails in English only), all non-English Fonts might be removed as well (and added after booting into the live env.)
Excessive Packages: Inkscape, Scribus, GIMP, Libre-Office (Abiword may be added after booting as an alternative), PiTiVi, …
The aim is clear: To be able to deploy Tails with the very limited resources that we have, and use it for communication as fast as possible.


Subtasks


Related issues

Related to Tails - Feature #15291: Remove less popular packages that users could install themselves Resolved 2018-04-17
Blocks Tails - Feature #15392: Core work 2018Q2 → 2018Q3: User experience Resolved 2018-03-09

History

#1 Updated by Anonymous 2018-06-04 09:46:14

  • Description updated

#2 Updated by Anonymous 2018-06-04 09:47:02

  • Type of work changed from Promote to Research

#3 Updated by mercedes508 2018-06-07 11:55:21

  • Type of work changed from Research to Discuss

#4 Updated by mercedes508 2018-06-10 17:31:22

  • Assignee set to sajolida
  • Affected tool deleted (Server)

Do you think it could be worth mentionning in the FAQ the reasons why we don’t provide such a light Tails version?

#5 Updated by Rover11 2018-06-11 05:56:53

Removing these packages from the original Tails live reduces it’s size by 465 MB without any impact on daily usage:

1) Inkscape (133 MB) (probably used by a minority of Tails users)
2) libpyzy (63 MB) (If The tails-light only supports English, all fonts can be removed)
3) Gimp (46+ MB) (EOG already provides the means to view photos, editing photos usually is used by fewer people)
4) Scribus (37+ MB) (probably used by a minority of Tails users)
5) Fonts-arphic-* (40 MB)
6) Fonts-unfonts-core & Fonts-wgy-zenhei (50 MB)

I am hesitant to add Gnome-user guide to this list, however, there can be an optimization of usability vs final ISO size. Also, there can be a poll to get an stat for the usage of standalone packages like audacity and pitivi to decide a cut-off for including apps in the Tails-light.

#6 Updated by Rover11 2018-06-11 06:07:15

mercedes508 wrote:
> Do you think it could be worth mentioning in the FAQ the reasons why we don’t provide such a light Tails version?

If this tread concludes that there shouldn’t be, it is good to know why Tails in it’s current state, has so many packages pre-installed.
essentially: how the first decision was made to prioritize packages for editing multimedia over applications like VeraCrypt, Gnu Social, Gnu Ring, and it’s likes.

#7 Updated by Gaff 2018-06-11 08:27:09

Presumably once the [url=https://tails.boum.org/blueprint/additional_software_packages/]Additional Software feature[/url] is working there will be a very compelling case for stripping all the multimedia packages from tails default?

#8 Updated by Gaff 2018-06-11 08:52:16

Presumably once the Additional Software feature is working there will be a very compelling case for stripping all the multimedia packages from tails default?

#9 Updated by intrigeri 2018-06-11 08:57:59

  • related to Feature #15291: Remove less popular packages that users could install themselves added

#10 Updated by intrigeri 2018-06-11 08:58:13

> Presumably once the Additional Software
> feature
is working
> there will be a very compelling case for stripping all the multimedia packages from
> tails default?

Yes, we’re considering removing at least some: Feature #15291.

#11 Updated by intrigeri 2018-07-03 10:55:01

  • Category deleted (Accessibility)
  • Target version deleted (Tails_4.0)
  • Estimated time deleted (6 h)

#12 Updated by intrigeri 2018-07-03 10:58:19

Rover11 wrote:
> essentially: how the first decision was made to prioritize packages for editing multimedia

Three things:

  • Start with our design doc (e.g. “2.6.3.3 Document production applications”).
  • Until recently, additional software packages did not work while offline, so the apps needed to support this use case had to be included by default.
  • Until Tails 3.9, additional software packages are unusable for most of our target userbase.

> over applications like VeraCrypt, Gnu Social, Gnu Ring, and it’s likes.

Last time I checked, VeraCrypt and GNU Social are not in Debian, and Ring did not work over Tor.

#13 Updated by intrigeri 2018-07-03 11:07:11

Rover11 wrote:
> Removing these packages from the original Tails live reduces it’s size by 465 MB

Does this really reduce the size of the ISO by 465MB? Or merely the size of the data on the root filesystem (uncompressed)?

Anyway, reading this ticket again, even if reducing the ISO size by 50% was easily feasible, and even without thinking about what part of our limited resources would need to be diverted towards this project to make it work, I’m not convinced:

  • wrt. initial installation: I guess that the set of locations where dividing download time by two is a game-changer is small and it won’t help the people who currently are impacted the most by the “Tails ISO is too big” problem as it’ll still be too big for them (I would expect that either downloading hundreds of MBs is doable, or it’s not)
  • wrt. upgrading: we have automatic (incremental) upgrades so “downloading Tails ISO for every release” should not be a thing; granted, we’ve messed this up a couple times this year, but we’re actively working on what’s needed to avoid breaking automatic upgrades again in the future; note that the size of automatic upgrades would not be impacted much by removing multimedia package (most of their size is caused by Tor Browser, Thunderbird and kernel upgrades)

#14 Updated by sajolida 2018-07-19 15:51:52

I fully agree with intrigeri here. As an experiment, I’m trying to even more apps in f4c21628f4 (Feature #15291#note-15).

#15 Updated by sajolida 2018-08-06 18:46:09

  • blocks Feature #15392: Core work 2018Q2 → 2018Q3: User experience added

#16 Updated by Anonymous 2018-08-16 10:56:06

  • Subject changed from Tails-core script to Make Tails ISO smaller

#17 Updated by Anonymous 2018-08-16 10:56:23

  • Subject changed from Make Tails ISO smaller to Consider making Tails ISO smaller by removing packages

#18 Updated by Anonymous 2018-08-16 10:56:34

  • Status changed from New to Confirmed

#19 Updated by sajolida 2018-08-17 08:21:47

  • Subject changed from Consider making Tails ISO smaller by removing packages to Consider distributing an alternative and minimal ISO image
  • Priority changed from Low to Normal

I’m moving this ticket back to: New. The idea behind this one is not to make the ISO smaller in general (that’s Feature #15291) but to distribute an alternative minimal ISO image.

I’ve been investigating on Feature #15291 whether it would make sense but it’s still not clear whether we want that. So I’m retitling the ticket to be more specific about the research and discussion about that.

#20 Updated by Anonymous 2018-08-17 09:49:17

Ack :)

#21 Updated by letthemeatpie 2018-09-11 03:16:15

IMO, this feature is a must have. Many of us don’t need all of the fat.

In fact, every time I launch Tails I run a list of packages for removal through Synaptic so I don’t have to see them on my system. What remains is bare bones. If I could easily re-master a small image myself for private use I’d do it. I’d even go so far as to replace Gnome with something much lighter. (Yes, I’m aware of the ‘heads’ project, but choose not to use it) If I could I’d even remove the ‘Cheese’ packages and many others, but they want to take out the Gnome Control Center. And others, well they want to remove vital Gnome components, but I really remove as much of the software as I can. It’s not a solution by any means, but it works for me.

#22 Updated by sajolida 2018-09-12 17:02:27

  • Status changed from Confirmed to Rejected
  • Assignee deleted (sajolida)

Reporting on the experiments that I did in Feature #15291. These are very rough numbers and very rough experiments on the ISO, just to see evaluate the idea of this ticket based on real ISO size.

devel 1168 MB
remove less popular packages 02535f38b9 1121 MB –4%
remove document editing f4c21628f4 943 MB –19%
remove fonts b7b0122d57 886 MB –24%
remove locales bda974fa5d 878 MB –25%

So stripping down Tails that much (English only, browser and mail only) would gain 25% of image size. That still wouldn’t fit on a CD.

I don’t think 25% of 1200 MB will make a real different in terms of adoption for new users.

I understand the point of removing some of the “fat” now that we have a better Additional Software, that’s Feature #15291 and I want to work on it. But I want to do that for everybody, not in an alternative image.

So, my experiment reenforced my position of thinking that it’s really not worth it distributing 2 different images. In other words, we could do so many more things with the little ressources that we have to make Tails useful for more people.

#23 Updated by intrigeri 2018-09-12 20:42:07

> I don’t think 25% of 1200 MB will make a real different in terms of adoption for new users.

> I understand the point of removing some of the “fat” now that we have a better Additional Software, that’s Feature #15291 and I want to work on it. But I want to do that for everybody, not in an alternative image.

> So, my experiment reenforced my position of thinking that it’s really not worth it distributing 2 different images. In other words, we could do so many more things with the little ressources that we have to make Tails useful for more people.

Fully agreed.

#24 Updated by letthemeatpie 2018-09-13 06:41:57

>So stripping down Tails that much (English only, browser and mail only) would gain 25% of image size. That still wouldn’t fit on a CD.

Well one could blame that on Gnome. :P

If one really stripped Tails down, including the removal of Gnome and replacement of a simple WM, it would likely fit on a CD. My concern for shipping a slim version wasn’t exactly for size, but to reduce the amount of packages for which many users of a slim version would never use. (and they could install additional as they wish) It also reduces the attack surface in reducing the amount of bugs to be exploited in many packages. Many programs, especially larger ones dealing with audio/video access/use of any kind, weren’t designed for such an environment. There are just too many unknowns. There are other benefits to ‘trimming the fat’ too but I’ll stop here. This idea is probably best left for someone else for their pet project, rather than Tails dev.

Thanks for reading, for the reply and the report of your experiments!

#25 Updated by sajolida 2018-09-13 17:24:37

> Well one could blame that on Gnome. :P

We could blame GNOME for being fat… or praise it for being the most
usable Linux desktop environment. The two facts are probably related but
Tails won’t ever use anything else than GNOME until there is a better
alternative in terms of UX (usability, integration, aesthetics, etc.)

> My concern for shipping a slim version wasn’t exactly for size, but to reduce the amount of packages for which many users of a slim version would never use. (and they could install additional as they wish) It also reduces the attack surface in reducing the amount of bugs to be exploited in many packages.

We agree on that and that’s what Feature #15291 is about.

> Thanks for reading, for the reply and the report of your experiments!

You’re welcome :)