🎹 Tim Janik ✅ @timj@social.tchncs.de

1984

The Assange case for extradition is still 90% upheld.
The judge refused extradition because the pending suicide risk would make it act of oppression.
The US going to appeal, so the decision will be taken to a higher court and has to be fought all over again.
The lawyers are now talking to Assange to prepare bail application, to be heared in 45 min.
That means, the US prision system is so bad that even this judge wouldn't send Julian into it.

via @johnwrees #FreeAssange

Just now:

Judge rules against US extradition of Julian Assange, contending extradition would be oppressive by reason of Assange’s mental health.

The judge and defense are discussing an application for bail. US government is going to immediately appeal.

via @kgosztola #FreeAssange

A very happy new year #2021 for everyone.
This year it is time for cleanup! ;-)

I'm porting the #Beast web UI from Vue-2 to #Vue-3. Turns out to be more involved than I originally anticipated.

I kept getting an odd exception during my tests. It looks like Vue-3.0.0 can become confused when state watchers (created via Vue.$watch) are unregistered during a later stage of a component lifecycle - trying to remove item from `null`.

I fail to see the error in my code at least (which works fine in Vue-2) and suspect it's a genuine Vue bug, filed here:

https://github.com/vuejs/vue-next/issues/2381

Has anyone experience with deterministically building #woff2 files for #reproduciblebuilds?

Using svgtofont via npm I've not managed to get this to work yet.

Grabbing a coffee, the #Debian conference starts in 5 minutes:
https://debconf20.debconf.org/

Now I'm wondering if Pen + Eraser work better as piano roll cursors, or if it is better to always use a cross-hair + tool icon cursor, like #Gimp does it.

Some users want to intentionally disable that though...
https://www.reddit.com/r/GIMP/comments/8zuvtf/how_to_get_rid_of_cross_symbol_over_paintbrush/

The standard CSS cursors don't really look suitable to convey insert + erase for piano roll notes.

Here is a first draft at creating cursors myself for editing #MIDI notes in the #beast piano roll editor, hotspots marked red.

Fiddling…

I'm working on a new knob for #Beast, with bidirectional and unidirectional modes.

The updates may only use "transform:rotate(angle)" #CSS to utilize #GPU acceleration during automation.

Sadly, Chrome cannot compose *inside* and SVG, so I have some nasty splitting and layering of the #SVG elements going on...

3/3 While the ability to run at nice priority -20 still requires root permission for the initial setup (or SYS_CAP_NICE), even that could *potentially* change with the introduction of a latency nice setting, if the kernel developers manage to implement unelevated latency re-nicing without opening up the gates for DOS attacks.

These developments are described in the LWN article 'The many faces of "latency nice"':

https://lwn.net/SubscriberLink/820659/d293ce7f2c05139f/

#linux #audio #lad

2/3 Not only is the risk of system lockups unwarranted to avoid audio drop outs, there is also no need for it to be taken.

Just prioritizing the audio rendering threads to run at nice level -20 is sufficient to prevent drop outs in low-latency setups, i.e. setups with <10ms round trip time. BTW, such setups require audio fragment sizes <= 128 samples (or "periods with <= 128 frames" in ALSA language).

#linux #audio #lad

1/3 On the topic of Linux low-latency audio setups…

Installing jackd (1 or 2) still requests "realtime priority" to minimize latency.

This project should really get its act together and stop demanding this kind of elevation from benign audiophile users.

Don't enable this! Running a process with realtime scheduling priority allows it lock up the system for good (and no code is bug free).

#linux #audio #lad

Resistance is fruitful: 😀
CANN rejects sale of .org registry to for-profit investor group

https://www.reuters.com/article/us-icann-org-sale-idUSKBN22D4FV
#SaveDotOrg

BAD G++!

Apparently, g++-9.2.1 does *not* const-initialize a `static Class x;` with a constexpr ctor, *if* it also happens to have an uninitialized uint64_t array member. It does not generate a warning either, but surely triggers the "C++ Static Initialization Order Fiasco" at some inconvenient future point.

#cplusplus #cpp