There is supposedly kernel-level mitigation for this now, although I haven't found it to work; the "weird issues" in this patch never went away so I'm abandoning it in favor of just using sysctl and a pkg-message.

This was entirely the wrong project / community / everything for this to land in.

This looks abandoned with 2 years of no movement. @michaelh do you still feel this is a WIP?

This looks abandoned with 2 years of no movement. @michaelh do you still feel this is a WIP?

This looks abandoned, no movement in over two years. @michaelh did you still want to introduce docId instead of quint64? This diff was supposed to supplant D10851 but seems to have fallen asleep as well. (In retrospect, perhaps I would have suggested first introducing just using docId = quint64; somewhere, to get the more-expressive-type-name part out of the way before real changes)

This looks abandoned, no movement in over two years. @michaelh do you want to explicitly abandon this one, so we have a clean slate for invent? None of the docId type changes seem to have made it to master (see also D10856)

"Overcome by events" as we say in FreeBSD-land

This was fixed independently in 114b349f09

Yes, sorry (I have no idea where Phab notifications go for me, so I missed the ping as well). These changes are good (and still apply).

Pushed via invent in 0c4ad7f1

If I mount the CD (e.g. by "make accessible from other applications") and then press the physical eject button (no applications are open on that mount), it does not eject. The disk stays mounted. I honestly don't know what behavior I expect, though.

Here's a screenshot of my device notifier. My machine has a CD-ROM drive. The drive contains an OpenBSD installer CD.

Bit of background, FreeBSD has HAL but is trying to get rid of it. It also has bsdisks, which is a FreeBSD implementation of the UDisks2 API -- this follows the "shim an appropriate API even if we don't like it" principle which was decided upon at the Devsummit at FOSDEM 2020. So in principle the Plasma (or Solid) code should just use the Linux API it needs from UDisks2 and FreeBSD will chase that.

Some new requests (I don't think @ognarb receives mail I send):

• a way to do 302 redirects (then I can tidy up the announcements with 302s instead of text, also better for search engines)
• I'd like to have the date in blog posts / news items pages, e.g. the list of news shows a date (floated right), but the news item itself, https://ev.kde.org/2020/02/24/kde-e-v-is-looking-for-marketing-support/ , does not show a date in the title.

I've done the merge now of news and announcements:

• announcements were the press releases, but we only ever did six or seven of them
• the associated news items just pointed to the press release

So I merged the content into news, put in some boilerplate text to send users to the news items from the announcement, and dropped the announcement item from the menu. The old announcements URLs are still valid, but tell the user to go to the news item. They're not linked directly.

• I don't know if it's doable, but ev.kde.org is a bit odd in that it's "owned" not by the community, but by the e.V. itself. The maintained-by: link at the bottom on the current site points to the board; the new one to www@ .
• The Information -> About Us link is kind of useless, since it goes back to the homepage. I don't know if there's an adequate modern-www paradigm for handling that. (This also applies to the current site)
• On What is KDE e.V. the blockquote from the articles of association doesn't look like a blockquote.
• The news item about the 2018 report has gone missing
• Can we merge News and Announcements? Escpecially now the page title for news is Announcements.

(These are non-blockers to a migration, and I'd fix them directly myself once the site is migrated)

I thought things went pretty well this year, without a great deal of coordination overhead.

I'll tell the xrdesktop folks we can wrangle some space for a small demo, to be figured out once we're there. Since we don't have the big Krita demo, we can do something else fun with the monitor.

The Krita showcase is a "maybe", since Boud isn't coming and Wolthera is a "maybe". The xrdesktop folks were asking if they could use a corner (well, enough space for their demo)

I have a 1mx1m KDE logo on canvas, suction cups to hang it with, and a rollup banner that we can use. I could try to get a blue tablecloth at the market tomorrow (last year we used Leinir's cloth, but that's not here now). @ognarb can you check those off on the list?

@ognarb did you manage to order merch? I've been totally useless here.

@ognarb No and no, respectively.

• Expand dynamic cast hierarchy. This is another situation where a 2-stage dynamic_cast<> doesn't work, leading to a nullptr dereference.

qgit, https://github.com/tibirna/qgit, (Tibirna is an old-time KDE contributor).

Calamares needs to be run as root, unfortunately (so that it can mangle disks and write essentially arbitrary files in the target filesystem). That is usually done through pkexec or sudo. In both cases, you need to take care to preserve enough of the environment that Calamares can find the theming. I'm not sure if pkexec -E is sufficient. In any case, by pointing enough XDG things to the home directory of the live user, rather than root, you can improve theming without doing any work on / in Calamares itself. Not resetting $HOME might be enough.

Now you (@davidedmundson ) mention it, yeah .. I've only tested positively, to see if konsole then does what I want. You're right that this could have surprising effects when multiple unique applications share a prefix (e.g. org.kde.plasma.something and org.kde.plasma.somethingentirelydifferent). There would have to be a closer check to find "this is really an instance of the intended application" and not have accidents with startsWith().

This has been applied to FreeBSD ports because it works. +1 (with the other konsole patch that is on invent).

(also change is obsolete and should be abandoned)

Commit ae4027d640d3297e78d8d0c87389aade128476a5 fixes the problem in a different way, by changing QList<T> parameters to QVector<T>.

Looked for -0 alternatives, couldn't find any. Dave tells me there's sufficient fallbacks, and otherwise we'll just patch it out in packaging.

Let's get this in, and then quibble about how much commentary needs to be written for the _p.h file (since, looking back, it's a bit heavy on the C++ fancyness -- my Opinions have changed a little)

(Assigned to Ivana because she had the FOSDEM 2019 one, but that shouldn't mean that this is her job)

• Discussion so far has happened on mailing lists (community, *Climate Impact and KDE*, ev-membership, *Environmental Responsibility Policy*).
• Draft policy was presented at https://share.kde.org/s/JN2bY6qbACbidYJ and has been polished and edited since.
• Some lobbying has started outside of these regular discussion channels as well.

LGTM on the FreeBSD side (I've checked, the procctl() code does block debugger access which is all we're asking to do).