HNHacker News
TopNewBestAskShowJobs

james_s_tayler

2,220 karma · joined October 30, 2018

submissionscomments
james_s_tayler··on The Linux codebase has over 3k TODO comments, many from over a decade ago
Every codebase is like this because it's cheap to open tickets and pepper the codebase with TODOs but orders of magnitude more expensive time-wise to actually do the work. The method of storing the information and prioritizing the work is irrelevant. You simply don't have enough bandwidth to do it all and some of it simply isn't materially important enough to ever get prioritized.
james_s_tayler··on The Linux codebase has over 3k TODO comments, many from over a decade ago
Open a ticket for that straight away and start working on it immediately. Or 3 years later. Whichever actually happens.
james_s_tayler··on Encoding your WiFi access point password into a QR code
>you have to trust Google to not store that URL anywhere

You also have to trust Google to still be operating the service by the time you come to use it.

james_s_tayler··on Where do individual contributors go once they peak?
So true. The reply is like "by these other huge famous companies have them too".

Ok. That's great. How about the small product companies that have like 20 or 200 engineers as opposed to 2,000 or 20,000?

james_s_tayler··on Functional programming design patterns (2015) [video]
This is exactly the resources I've been looking for to advance my understanding of how to accomplish certain things in FP before trying to learn it wholesale.

I have his book Domain Modelling Made Functional as it comes very highly recommended. I haven't gotten to it yet, but I'm even more excited for it now after watching this.

james_s_tayler··on Nim vs. Crystal
>We’ve been waiting for years

Spin up a Linux VM and get developing?

I do .NET for my $DAYJOB and we use Windows for development. It works, but I wouldn't constrain myself only to Windows for development. I have an Ubuntu VM at home that I do dev in for other stuff.

james_s_tayler··on Ask HN: Haven't worked for a while, best guide/advice to start a hobby project?
Something I have found really great is to purposely build something you will never ever even be remotely tempted to ship but that has some degree of complexity to it. What's great about this approach is you are never forced to be bogged down in details that aren't relevant to you.

If you ship a product and I mean like really try to ship it with a launch and everything, you will spend a tonne of time on a) non-technical tasks that don't matter to your day job and b) technical tasks that don't matter to your day job.

For example graphic design, copywriting, marketing etc. All great stuff if you want to go the route of actually trying to launch a company. Terrible to get caught up in if you just want to learn to write production grade systems.

Technical tasks you have to spend time on for stuff you actually will ship do have some value but really it can be limited. You might spend time to build deployment pipelines or auth or HA deployments etc. All of that stuff is great IMHO and I personally value having done a lot of it but to a lot of engineers who work on established code bases and just never ever touch that stuff I feel like it's not a huge value add.

With a project you know you'll never ship you can focus only on the parts you want to. You can then layer lots of little mini-projects into it and use it to try out new things over time.

I guess I'd recommend trying to ship a real project solo at least once too. It gives you perspective.

james_s_tayler··on The World of Statecharts
That's ok. Anything publicly available you used as reference materials? I had a read through that site, but id love to see what's considered a good example in code.
james_s_tayler··on An IT migration corrupted 1.3B customer records
The problem with the big bang rewrite is it violates Gall's law

>All complex systems that work evolved from simpler systems that worked

james_s_tayler··on How to undo almost anything with Git (2015)
Can confirm.
james_s_tayler··on How to undo almost anything with Git (2015)
Like others comment it's possible to scrub it from the repo entirely but this will change every commit hash (I'm guessing back to the commit where the file first appeared). It's probably something you need to coordinate carefully among all the teams. Have no open branches, scrub it, force push that then people can resume work.
james_s_tayler··on The World of Statecharts
This sounds very interesting! Do you have a link to a repo or tutorial / good reference example?
james_s_tayler··on Show HN: Mandarin Tone Trainer – A ‘Helping Ear’ for Mandarin Pronunciation
You do both at the same time if you want the fastest results. You get structured i+1 input that you can cope with for part of your learning and then you dive deep into the the native content and let your brain compute the statistics necessary to eventually be able to recognize the patterns at natural speed.

My definition is of listening comprehension is any information I can glean from listening. Information being a reduction in uncertainty. Doesn't need to be complete. Progress is directional.

I know my language learning algorithm well and I pound it on repeat until my goal is achieved.

james_s_tayler··on Show HN: Mandarin Tone Trainer – A ‘Helping Ear’ for Mandarin Pronunciation
Exactly. Hence the suggestion. Just watch the Taiwanese stuff.

I've found close to nothing culturally worthwhile from the mainland.

Edit: actually, now that I think about it... I could never find anything when I was outside the mainland but I went there for a 3 month business trip one time and I signed up to various local services (Spotify / Netflix equivalents) and there f;r the first time I felt I had access to a decent amount of content I enjoyed.

james_s_tayler··on Show HN: Mandarin Tone Trainer – A ‘Helping Ear’ for Mandarin Pronunciation
There are some alright dramas that come out of Taiwan.
james_s_tayler··on Show HN: Mandarin Tone Trainer – A ‘Helping Ear’ for Mandarin Pronunciation
>Front-loading explicit study of grammar is a total waste of time. Memorizing atomized vocabulary words is also relatively ineffective.

Self studied Japanese to fluency that way. AJATT method but largely concentrated on front loading all the grammar in 4 months then built 10k vocab over the following year. Worked like an absolute charm.

james_s_tayler··on Show HN: Mandarin Tone Trainer – A ‘Helping Ear’ for Mandarin Pronunciation
>Listening comprehension, IMHO is the last skill to develop in learning a foreign language. The reason is that it's "all or nothing". You either know and recognize all of the words in a sentence, or you can't cope with the sentence at all

So not true. Maybe depends on the way you study but I focus on listening skills hardcore early on. It's what helps build the natural accent. It's definitely not all or nothing. There are islands of understanding. They tend to grow and grow. It starts out with single words, then grows to two words together then grows to partial sentences and then to full sentences among a paragraph. So on and so forth.

james_s_tayler··on When an SQL Database Makes a Great Pub/Sub
Sure it is. If one approach is easier to mess up than the other then take the approach with less footguns.
james_s_tayler··on When an SQL Database Makes a Great Pub/Sub
The key thing about the performance aspect is SQL produces bi-modal behavior due to there being a query planner / operator execution. If all of a sudden your performance characteristics change dramatically because the query planner produced a different result than normal due to something changing about your dataset, that could be pretty bad. So, not having this risk of bi-modal behavior is a nice feature of dedicated pub/sub and queue software.
james_s_tayler··on How Shopify Manages API Versioning and Breaking Changes
Where I work, we have found 3rd Party SDKs to be an anti-pattern for this and other reasons.
james_s_tayler··on How Shopify Manages API Versioning and Breaking Changes
Magento makes me shudder.
james_s_tayler··on To be a good software engineer, become a French skeptic
I agree. Even for a comment on HN it was pretty vapid.
james_s_tayler··on A hiring test I'd like to run
There's a big difference between spending 50 minutes fiddling with that stuff when you're employed and evaluating it as part of the research phase for a project vs. in an interview with a gun to your head.
james_s_tayler··on Bitmain Rival’s Founder Arrested as War of Crypto Miners Heats up
Good while it lasts though.
james_s_tayler··on Bitmain Rival’s Founder Arrested as War of Crypto Miners Heats up
Unless you're the despot. In which case they are the best place to live.
james_s_tayler··on Google Brass Set 2023 as Deadline to Beat Amazon, Microsoft in Cloud
Also AWS is just the best at marketing their services. They did what Google did with search. We always say we "Google" things.

Developers these days say "Chuck it in S3" or "Run it on EC2". I feel they have a strong advantage because they've kind of come to own the language of the cloud.

james_s_tayler··on Ask HN: Books you plan to read in 2020?
Do you have a good example of a reference architecture or implementation of a non-trivial system you can point me to?

I've recently gotten a copy of the book Domain Modelling Made Functional. I'm yet to read it, but I'm pretty intrigued by some of what I've seen and I have heard it comes highly recommend.

I'd be very keen to hear of other high quality examples of things people consider good reference material.

james_s_tayler··on Ask HN: Books you plan to read in 2020?
Thanks. I've observed this as a general trend too. I'm starting to become more and more interested in the functional way of thinking.

The image you linked it interesting. I want to dive more in on the how/specifics. I don't necessarily buy the argument it's trying to make that "just make a function and you no longer have a need for factory pattern". You're trying to accomplish something fairly specific with that. Functions compose, I could see how there are properties they have that can be leveraged to achieve the same goal, but I'm guessing the functions you write or the way you write functions to achieve each of those specific aims likely differ in subtle yet important ways. I'm very interested in those details.

james_s_tayler··on Ask HN: Books you plan to read in 2020?
Meditations has some cool phrases here and there. It mostly gets very repetitive and you figure out his philosophy pretty clearly early on because he restates the same idea in hundreds of different ways. It's largely the same idea though.
james_s_tayler··on Ask HN: Books you plan to read in 2020?
What has replaced them?

Presumably some of the problems they intend to solve still exist.

Is there some resource you can point to that says in the past when faced with situation X developers tended to use pattern Y, these days that has been superseded with way of doing things Z?

I've seen some of those patterns leveraged to great effect in modern software. But genuinely curious about alternative ways to solve the same problems.

I've definitely noticed the trend towards more and more functional programming these days. I doubt it's simply a case of "just use FP and there are no more problems".

← PreviousPage 13 of 34Next →