dick applebaum

I think that the Home Server/HomeKit/Siri issues can be resolved!

Consider, If you had a Home Server, likely, you would:

• connect HomeKit devices via WiFi with a bridge if needed
• percolate up/trickle down cloud data and buffer locally LRU -- as iTunes interfaces iCloud files on the Mac over WiFi
• access local data from various iDevices and Macs via WiFi -- same as above, except local files only
• control the whole thing with Siri over WiFi - with Siri capabilities processed locally and tailored to your use off Siri

So, where do these local data reside, and what does this Home Server Box look like (tech specs).   Simply stated, it's [what's a] computer with some storage, WiFi, and voice recognition.

The Home Server can be a logical implementation rather than a physical one. It can run concurrently, on multiple iPhones, iPads, Macs -- all connected by Wifi and controlled by local Smart Siri.

Smart Siri will know about all your data, where they reside, and if they are currently accessible -- and be at your service!

Think about it!

Swift 5:

• stable ABI
• @dynamicMemberLookup
• FoundationDB API

Soli said:

1) "tech vs hubris"? Why do you think these are at odds with other?

I don't think that these are at odds -- its just that you and several others have been arguing over them for the last 20, or so, posts -- and missing the big picture -- the users decide -- not some techies on a manufacturer-centric forum.

All this sturm und drang about tech vs hubris misses at least two important points:

1. the User Experience
2. the users' trust  that the manufacturer has the means to provide and support it over time

ascii said:

I would like to see Apple implement client-side speech recognition. After all they are a client-side focussed company and claim to use "on device processing wherever possible" in their privacy literature.

Then they could make ads claiming that with HomePod, "No recordings are *ever* sent," which would be a clear differentiator from other products.

I believe that Apple has the tech to do that!  After all, they currently do client-side facial-recognition (point-cloud processing).

After playing around a bit with Apple's FoundationDB, I think that it can be a very fast and efficient embedded system on ARM -- and this could solve the local speech db needs.

Also, some of the algorithms used by Apple's Shazam could be used.

Finally, this client-side speech recognition could be implemented in stages:

• first, start with a local WiFi network to shared Macs or ARM devices
• ultimately, imbedded on the target device