GitLab

Port otr-bot to Python 3 and migrate to slixmpp

Closes #17031

See merge request !278

Test suite: bump timeout for the Greeter's GUI to update after language change.

See merge request !279

This reverts commit 4f2a8e52.

That approach worked fine on my system but not at all on Jenkins, so
not so robust after all. :)

Port Tails Installer to Python 3

Closes #10085 and #15010

See merge request !270

... when waiting for the Greeter's UI to settle after changing
language.

Otherwise it will spam our pretty log! It's even too verbose for debug
logging, imho.

It's particularly when switching to Chinese that the GUI takes time to
update, and while the 3 second sleep works on my system it seems
Jenkins is too slow, as it consistently fails in a way consistent with
this sleep being to short.

That bug (which actually was a libblockdev one) is fixed in Buster.

According to #10987 (comment 59103)
this was probably solved in Tails 3.0 (Stretch).

We have language negotiation on our website, that ensures the user is served the
localized page if there's one. But

Making these URLs translatable suggests translators they should hard-code the
URL to the localized page (e.g. install/download.fr.html), for which we don't
set up web server rewrite rules when we rename pages. In the past, this has lead
to such links being properly redirected in English, but yielding a 404 in
other languages.

Enable Electrum's Ledger hardware wallet support

Closes #15353

See merge request !272

Drop superfluous block.events_dfl_poll_msecs=1000 kernel parameter

See merge request !274

We really mean MiB = mebibytes here.

Detect corrupt GnuPG public keyring and restore its backup

Closes #17807

See merge request !275

Now flake8 is happy!

I dropped all but the 'ping' command, since that's the only one we're
using.

Closes: #17031

Otherwise, if we have drives sda1 and sda2 with same parent sda, only
the drive in the first iteration will have parent_data set since its
deleted in subsequent iterations.

We don't need to enable in-kernel media presence polling ourselves anymore:
systemd has been setting this to 2000 ms by default since v174 (2011!) and newer
.

refs #15353

Test suite: update expected title of the GitLab page we use

See merge request !271

Update deb.torproject.org's APT key

Closes #18042

See merge request !269

This reverts commit 29824dda.

So far I have only been testing it in a live session.

The purpose of this package is to keep deb.torproject.org's APT key up-to-date,
which does not work in Tails as we discourage users from updating packages.

On the build/dev side, this package only adds complexity and confusion: we have
to install the APT key ourselves anyway (otherwise live-build would not be able
to install packages we install from deb.torproject.org, before this keyring
package is installed). So this only adds a second place where we need to
maintain the same information, with no benefit. For example, I updated the key
in bfcd3b23, but the deb.torproject.org-keyring
package from the APT snapshot we're using replaces it, in the built images, with
an older, expired one.

It was removed in the initial porting to Python 3 for some reason, but
without it we'll encounter permission problems, so it is needed.

Apparently it used to work to pass a bytes-string of the whole
command (incl. arguments) to Popen() even with shell=False, but in
Python 3 it will interpret the whole string as the executable and thus
fails with "file not found" if any argument is given.

I imported the following:

    https://gitlab.com/saschamarkus/liveusb-creator/tree/feature/10085-Port-Tails-Installer-To-Python3

I gave up my attempt to import the commits one-by-one due to diverging
history causing quite a bit of conflicts, which was easier to deal
with in one go.

Thanks so much for your contribution, saschamarkus! Sorry that it took
so long! :S

refs #18042

Require Buster or newer for running our test suite

Closes #17842

See merge request !254