Revenue is the small part of it though. Every training and inference script in the ecosystem imports their client library and pulls from their hub by default, so they sit on the default path between a model and the machine that runs it. For company that sells the machines, owning that default is worth more than any subscription line.

Cheapest version of that review is already sitting in the format. Frontmatter carries created and updated, so you can sort by staleness and drop whatever nothing touched in months

Problem is not that nobody wants to do it, it is that GPU stack is exactly the kind of enormous driver surface Qubes exists to keep away from dom0. Second monitor does not really change that, somebody still has to trust the driver.

The driver could run in a VM and only display applications from that VM or a subset of VMs.

Most GPU drivers are an order of magnitude more complex than a CPU kernel.

There is also the fact most GPU are a bodged on architecture, and fundamentally impossible to really secure in the case of CUDA rootkits.

Fast, but a bad design. This is why we can't have nice things. =3

Main practical difference is that you can put index on generated column, and on view you cannot, SQLite has no materialized views. Constraint part you understood right, NOT NULL on generated column fails at insert, and note that VIRTUAL costs nothing on disk but can still be indexed, so STORED is mostly for when expression itself is expensive.

feedback can be wrong, so improver should check criticism against domain instead of writing it straight into skill, otherwise one reviewer bad taste becomes permanent rule for everybody.

It is real but not so dramatic as you heard. Offline routing must walk the graph across many downloaded regions, so time grows fast with distance. Trip inside one country is usually seconds, whole continent can be minutes. Standard trick is to put two or three intermediate points on the route. Then engine solves several short problems instead of one huge one, and answer comes almost immediately. Also OsmAnd can use online routing when you have signal, offline engine is fallback. So the bad case happens only when you plan very long route and have no connection at same time, which is not typical situation.

Everybody here compares CoMaps with OsmAnd, but for me most interesting part of article is different. Their bottleneck was not rendering or routing, it was that every map update needed new app release, so it had to pass through app store review. They decoupled this, processing went from 10 days to about 3, now fresh data weekly for whole planet, and around 35 hours when emergency is declared.

Very unglamorous engineering, and it is what actually made buildings traced in Tasking Manager reach people standing in the rubble.

Also worth to say: you can help with this part from home. During activation the tracing is done from satellite imagery, local knowledge is not required.