I am using Librera Reader. Make sure to install the F-Droid version (Librera FD) without Google Play services.
It is primarily an ebook reader that supports a variety of formats, but it is also an excellent PDF viewer. Significantly more feature-rich than any other FOSS PDF viewer for Android that I have found.
I think you are reading too much into this. SpaceX has a rather aggressive test program, and the purpose of the tests is not (only) to verify the functionality of the system, but also to learn about the vehicle and involved technologies. At this stage they are pushing the boundaries of the systems capabilities intentionally up to the point where it might start to fail. So if things don’t fail, it only means that they could have pushed harder and squeezed out more performance.
That does not mean that anyone wants Starship to explode. But the objective is clearly not to do a perfect launch, because SpaceX knows that it’s more efficient to make mistakes a few times times and then succeed, instead of spending excessive amounts of time and money one single perfect test launch. This has been communicated very clearly from the very beginning.
This sumarry is erroneously mixing up events from the first and second flight test.
I was confused for a second asking myself when Kroger was split into Kroger Sued and Kroger Nord.
Scott Manley conjectures that the staging itself worked as intended, but the flip maneuver caused too much stress on the internal plumbing (with all the propellant sloshing around), resulting in a ruptured pipe and burning in the base of the booster.
Not sure where you got that impression from. The booster apparently worked flawlessly during ascent, with all 33 engines working until stage separation. Only 30s into the boostback burn, the AFTS triggered after several engines shut down prematurely. The second stage was deployed successfully, and completed most of its burn, before being lost only around half a minute before scheduled (sub)orbit insertion, around 3km/h short of the required velocity.
For people like me without an X/Twitter account, there is also the spacex.com livestream (which is just the Twitter livestream embedded, but no login is required).
Hmm, I am a bit more confident in the ships abilities (at least for anything between hot staging and SECO).
Anyways, if it gets to the point to initiate hot staging (regardless of the outcome) and the FTS works, it‘s a success. But we should also remember that SN9 landed (crashed) harder than SN8, and SN12 was way worse than the previous three tests. If stage zero is mostly unharmed, the FTS works, and the authorities are not too unhappy, SpaceX has already produced enough hardware for several tests to get it right within the next few months. A good test is a test where you learn a lot, and can try again soon.
Here, the developer explained why development activity decreased:
While it is true that due to private reasons I had to take a bit of a pause of developing FlorisBoard and some time passed with no progress at all, implementing a completely new statistical NLP (Natural Language Processing) provider, or in laymans terms the long-awaited word prediction and spell-checking implementation, is also a huge task which takes a lot of time and trial-error and development time.
Android supports multiple payment providers. Some banks implement their own payment provider (e.g. Sparkasse in Germany), most just rely on Google Pay (now Google Wallet). Google Wallet has strict requirements for the Play Integrity API. Because of the modifications to Android that GrapheneOS is implementing, it is not eligible to receive the required integrity attestation and thus, Google Wallet is refusing to work. Google could at any point reconsider and certify/whitlelist GrapheneOS, which would allow Google Wallet to work using GrapheneOS. Likelihood close to 0.
Any banking app implementing their own payment provider is completely independent of this decision unless it also relies on Play Integrity API attestation (or a similar mechanism).
You are comparing GitLab (the application) with codeberg.org (the website operated by the codeberg e.V. non-profit). A fair comparison would be gitlab.com (the website operated by GitLab Inc.) with codeberg.org or GitLab (Community Edition or Enterprise Edition) with Forgejo (the application powering codeberg.org). They can be fully self-hosted and are both planning to implement AcivityPub-based federation.
The sidebar on his Patreon page says
you can actually donate any amount you want per month.
On this page, you can scroll down to the bottom, and there is an option “Or choose your own price” with a “Make custom pledge” button.
You might have heard of alternative frontends to services such as YouTube (Piped, Invidious), X/Twitter (Nitter), etc.
Something similar exists for Fandom: Breezewiki. This instance seems to occasionally result in errors, other instances seem to work more reliable.
You can request a copy of your data according to Art. 15 of the GDPR. If they reject this, escalate to your local data protection authority. This way, you could gain access to the data even if Google tries to block you.
… well, theoretically, at least.
We can only guess. But they can probably detect contacts for which the phone number is updated or which have several assigned phone numbers.
I would assume that “market share” is related to the relative number of units sold/number of active subscriptions/fraction of total sales in terms in revenue, or some similar metric. I run a variety of different distributions on servers (bare metal, VMs and containers) and desktop computers. Do they all count equally? Without giving it more thought, I wouldn’t even know how to determine the market share of Ubuntu in my own home in a sensible way.
With Windows, I can just count the number of active licenses. Oh wait, its zero.
This feels more like a hack than a suitable solution. Also, it does not work with static binaries.
It would be nice to have better support for 464xlat clat on linux. Sure, there are implementations like clatd based on tayga (which, by the way, was last updated in 2011???) or jool (out-of-tree kernel module), but setting them up is rather cumbersome. On Android, I automatically get a v4-rmnet0 interface when I connect to an IPv6 only network (such as my mobile network), which provides a local IPv4 gateway for legacy applications and websites, that will automatically do nat64/clat. Prefix-discovery is done using DNS64.
This is just a frontend, DeepL (or any other engine you use) still sees the text that you translate.
Me, a physics educator: “Interesting!” takes note