or Connect
New Posts  All Forums:

Posts by Dick Applebaum

I wonder if they are using FoundationDB as the supporting db -- at least for the Apple Online Store functions. A FoundationDB implementation would allow Apple to: add new products to the Online Store without taking it down flip a switch and cutover to the new products in seconds immediately scale to processing thousands (millions ?) of transactions per second avoid the customer frustration of having to start over due to hangs, crashes and timeouts (losing any items...
Finally! I'm more interested in the latest Xcode/Swift beta than iOS9 or Watch OS2
It's been my experience at the Apple Store, if you put something of high demand in your cart the quoted shipping date is honored -- if the order completes. From that, I imply they are reserving items until the transaction completes or times out.I can see a finite quantity in $ available to loan, Stocks available to sell, etc. Wouldn't the dependencies be checked before the transaction item reserved ... "You don't have enough funds in your brokerage account to buy 10,000...
FWIW, some digital transactions can encounter similar problems to those involving physical items -- where there are a finite amount of resources available. One classic example is Class Scheduling. Here's a good tutorial that shows how to handle transactional issues in Class Scheduling using FoundationDB in a Python app: https://foundationdb.com/key-value-store/documentation/class-scheduling.html I've never used Python or FoundationDB -- but I can understand what's...
I'm talking about lots (thousands, millions?) of customers, concurrently ordering multiple items -- some of which are in high demand. My most recent experience was getting up at midnight to try and order 6 Apple watches (1 for each family member and 1 for development) from the Apple online store. The app kept crashing/hanging/timing out -- i'd lose what was in my cart and view the ship times grow before my eyes. I eventually placed 6 separate orders for models/bands...
Man, you made me chuckle:  It depends on the implementation -- I don't know how many times I used those same words when I worked for IBM.From the way you answered the question, I assume that a modern SQL DB does nothing to detect or resolve the situation. If that is true, then both customer's transactions, in my example, will fail (timeout) with possible locks in place and the DB will have quantities reserved that were not sold. This will require some application code...
I can't answer that question as I have no ASP.NET Experience and very little Java experience. When Swift was announced, it was presented as an amalgam of the best capabilities of many modern languages implemented to provide code safety, performance, conciseness (readability/maintainability), etc. There must be a need for for some of this because of the dramatic rise in popularity of Swift in a very short time.
First the amount of data transmitted — on mobile it is very expensive -- LTE Data rates. HTML is quite inefficient, JavaScript is the pits. XML is about 70-90% overhead … JSON is better, but can easily be 30-50% overhead. So, if you have an app running on a mobile device, say Safari, that uses the LLVM -- why couldn't the web servers take all the crap they dynamically and convert it to LLVM code, download the smaller packet, then execute it on the device. Radical, I...
Sorry, I've been tag-teaming my computer with my grandson who's making a movie -- Beta El Capitan doesn't help!.I know that FoundationDB implemented an SQL Layer (SQL Interface) above their hash DB.Again, it’s been a while since I did any heave SQL programming (MS & Sybase). Has SQL ever resolved the fatal embrace issue::Part A has a qty of 1 in stockPart B has a qty of 1 in stockCustomer 1 orders (reserves) the last Part BCustomer 2 orders (reserves) the last Part...
True, but SQL (and its variants) are not necessarily the best data solution for all needs. That's my point.
New Posts  All Forums: