Jerry Zhao (@codingsince1985) profile picture

Craft business values in simplest way that makes customers (bottom line), developers (easy to digest) and infrastructure (CPU, RAM, I/O, etc) happy altogether.

Melbourne, Victoria
Joined August 2014
Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)Photo by Jerry Zhao (@codingsince1985)
0) when no microservices, return over cost. 1) if you get nothing from having 10 microservices, same return over 10x cost. 2) if you get something (say 20x) from having 10 microservices, sometimes you can reduce (say 5x) cost by using circuit breaker, so 20x return over 5x cost.
Resilience math: If your system depends on 10 microservices that all have 99.9 uptime, your total uptime will only be 99.0. Literally 10x worse (88h v 8h downtime pr year).
71
as a software developer, what makes more sense to me is, multi-core benchmark / number of threads. for example, although Core i7-1280P (20 threads) and Ryzen 7 6800U (16 threads) have close multi-core benchmarks, your application will more likely perform better on latter.
1
67
i'm assuming your application is developed in non-blocking / asynchronous way. otherwise, no benchmark makes any sense to your application's performance. #Software #Benchmark #CPU
1
25
for multi-threading cpu intensive tasks, i would further make it multi-core benchmark / number of cores. 2nd thread on a p-core is even slower than the only thread on an e-core.
17
almost same here. 29 years ago, i wrote detailed design, C and COBOL code in vi (not vim) from a VT100 terminal.
Thirty years ago I _lived_ in EMacs. Everything I did, I did in EMacs. I read and wrote email, I posted on UseNet, I perused directories, I edited, compiled, and debugged. I just left EMacs running all the time. It was the only thing on my screen. Then came the IDEs. I was smitten by IntelliJ. (I still am). So I left EMacs for the more sophisticated GUI of IDEs. Now I wonder about that choice. My toe is back in the Emacs waters and I can see some advantages. I can also see some costs. I tread carefully.
35
$ sudo fstrim -v / /: 28.9 GiB (31058874368 bytes) trimmed #nye
Replying to @codingsince1985
$ sudo fstrim -v / /: 1.5 GiB (1630089216 bytes) trimmed #nye
15
#JuliaLang quiz today #Julia言語 which one is true if a = (3,2,1)?
33%sort(a) returns (1,2,3)
0%sort!(a) returns (1,2,3)
33%both 1 and 2
33%neither 1 or 2
3 votes • Final results
1
167
#JuliaLang quiz today #Julia言語 m = [1 2; 3 4; 5 6] (a, b) = m what's in b?
18%2
35%3
12%[3 4; 5 6]
35%error occurred
17 votes • Final results
1
2
1
2
1,140
35.3% got it correct.
24
new build-in functions min and max simplify solution to 78 algorithm problems by eliminating 578 LoC. well done, @golang team. #golang github.com/codingsince1985/U…
20
those who jump ship frequently don't care the other 20%, which is not only hard to master, but also not portable when jumping to next tech stack's 80%. hello world all the way.
Learning a new tech stack is an 80/20 thing. You can learn 80% in a relatively short period. (2-3 weeks depending on effort and immersion). That’s enough to be productive and effective. The other 20% takes significantly longer and provides far less benefit.
24
i'm not sure if that's the case for hiring, or for planning a system. however it's true that more than enough effort is spent on choosing language / platform. in java world, there're few things after _last_ things, like gradle vs maven, kotlin vs groovy for gradle script, etc.
To whom it may concern: When hiring programmers, language and platform are the _last_ things you should consider.
74
old devices start degrade in performance when getting new system / firmware that was developed with only newer devices (which may be more than 2x powerful) in mind. i have this device, and google did a good job not making it unusable over the years. thank you. #google #chromecast
Google has quietly ended support for the original Chromecast. 10 years wasn't a bad run, and you may be able to continue using it. Just don't expect software or security updates, and "users may notice a degradation in performance." buff.ly/3OODCtw
53
$ sudo fstrim -v / /: 88.4 GiB (94940233728 bytes) trimmed #nye
$ sudo fstrim -v / /: 13.7 GiB (14687219712 bytes) trimmed #nye
1
$ sudo fstrim -v / /: 1.5 GiB (1630089216 bytes) trimmed #nye
1
29
Moore's Law is not as relevant any more, considering the size of recent chips and the power required to drive those chips. What we expected from Moore's Law is actually, computational progress will become significantly faster, smaller, and more efficient over time.
in my 35+ years of coding, i’ve learned that… now i need to prove i know something.
In my 25+ years of doing software, i’ve learned that… I don’t know anything
well, that's a lot i reckon. however what's the RDI (reference daily intake) for battery? @OriginalBonsoy , @AmazonAustralia amazon.com.au/dp/B004JBV88E
or i should have asked @healthstarsau / @healthgovau if #HealthStars / #HealthStarRating is broken.
i've no idea how @ALDIAustralia made this possible. i reckon it must be very hard.