Blog‎ > ‎

STBC, Facial Recognition, Digital Citizenship, ECMAScript 2015, AsyncIO, Coroutines, X-Road, e-receipts

posted Dec 24, 2015, 11:10 PM by Sami Lehtinen   [ updated Dec 24, 2015, 11:10 PM ]
  • Reminded my self about Space-time block code (STBC)
  • One directory got so many files that it was really annoying to access. Because files were hex hash named, I made the same trick which GIT uses. First two characters of hex code are now used as subdirectory name and rest of hex is used for file name. So instead of having 100k files named like 12345678 in one directory, now we got only less than 4000 files per directory and 256 directories from 00 to FF. Now it's much nicer to handle those files. If number grows even larger, I could add one layer. Btw. Squid cache used to use this ages ago. You could choose if you use 1-2 hash item buckets and how many nested buckets you're going to use. like /1/2/3/45678 if you did choose three layers of one character. Or maybe two layers with two characters like /12/34/5678.
  • Facial Recognition will be big stuff in future of customer loyalty and security business. When you walk in store, the staff will be alerted. Are you good customer? Do you prefer cheap or premium stuff. Which customers should be served and so on. System will automatically also alert if you're unwanted person etc. Casinos have had this tech for quite a while. But soon it will be standard feature on video surveillance systems which are already everywhere. At the same time, very cool yet extremely scary. Paying in cash only, won't make you anonymous any more.
  • CloudFlare POP map shows Copenhagen located at Oslo, Norway. Hmm, they're clearly having way too many POPs because they don't seem to know where those actually are. - Yes, they've already fixed that.
  • Digital Citizenship is interesting concept. I wonder if it's any different using any online authentication system. It's just online identity which you can use with official authorities. No big deal. Yet identity & identification market got very high diversity and there aren't many widely used solutions. It's horrible to say, but maybe Facebook is one of the best identity management systems in practical sense. How could I say something horrible like that? Well, because it really doesn't matter if it's technically perfect, if nobody's using it.
  • Studied Materialize CSS in more detail. There are a few bugs and kinks which I don't really like. Like with jQuery Mobile you at times needed up fighting with the framework and even with basic settings some things just didn't work out. yet jQuery mobile seemd to work much better than the Materialize CSS. On mobile and with different browsers there are all kind of minor annoying issues. Like things getting over text, or text getting under image frame or something similar.
  • Long nice article about JavaScript (ECMAScript 2015) Promises and programming.
  • I was just playing with asyncio, when I encountered a strange problem. I just couldn't find ensure_future from my asyncio lib, even if I was checking Python 3.4.3 documentation. After a while I noticed that the documentation said that new in version 3.4.4. Nice. I did somehow expect that documentation for 3.4.3 wouldn't include future features. Pun intended. I guess the reason for this is that there's isn't separate documentation for every version released.
  • Went through Python 3.4. asyncio and coroutines again. Yet unfortunately I didn't find anything I could especially user right now. Coroutines are suitable for only certain use cases.
  • One guy used to write all emails in Gmail and then then encrypt those using PGP and send. Without realizing that the unencrypted draft is stored too. So much fail! 
  • Finnish National Service Channel (X-Road, Enterprise Service Bus) is finally open and in production. That's great news.
  • Had very long discussions with many people about electronic receipts (e-receipt). Some silly people think that PDF send via email is e-receipt, no it isn't. It's not properly structured data. There needs to be common standard for processing all that data. Currently there's proposal based on Finvoice XML invoicing standard. The proposal expands the Finvoice with new fields required for receipt information. Using e-receipt would almost fully automatic travel expense tracking. Making travel expense reports and invoices is very common pain point for many organizations. The only remaining issue is how to identify the customer. As said earlier in many posts, this is the problem that nobody has solved so far. If customers can be identified secondary question is how to route and deliver the data. But that can be arranged quite easily compared to the first problem. Third problem is creating require incentive for organizations to invest in this technology. But I assume many customers would find automatic travel expense filing so awesome that there would be a big push from customers for businesses to adopt it. After all automati XBRL bookkeeping would be so nice when you travel.