4 ms·
This would mean hiring a total of six experienced developers. Let’s ignore the madenning tediousness of having to make every minor change six times on six
by Jtsummers 9y ago
This would mean hiring a total of six experienced
developers. Let’s ignore the madenning tediousness of
having to make every minor change six times on six
different platforms, and focus instead on the costs.
With an average salary of $150k (probably more for the
hard to find like Mac developers), and ignoring the
massive cost of finding and hiring these developers,
that’s a total of $900,000 in development costs every
year.
Focusing on the first part: Why would you be making the same changes 6 times? Is your code so tightly coupled between logic and display that you haven't reduced the common features to, essentially, libraries at this point? Separation of concerns is not a novel concept.
And Mac developers hard to find? From what I keep reading here, MacBook Pros are the development machine to have. I can't imagine that it's so hard to find a Mac developer as this post suggests.
As a one-person team deploying on all these platforms,
even the most minor change will take at minimum three
development days, one for each codebase.
If you're one, solitary, person, you aren't a team. Just like there's no army of one, despite the pervasive recruiting posters and videos years ago.
I'm glad Electron is letting you produce yet another note taking app. We need more of those. I hope it goes well. But please, please, please: I'm tired of my battery dying before I finish one cup of coffee at the coffee shop because of these inefficient apps.
Will someone, please, make Positron, the anti-Electron. It needs to be as portable as Electron (ideally), but based on a better runtime and language. Hell, Elixir is popular, let's go with that. You'll get great concurrency from the start, and hot code loading, and a decent language
- agentultra 9y agoMaybe https://github.com/Widdershin/positron https://github.com/Widdershin/positron A real contender to Electron is going to have to ship a runtime or libraries that support the DOM and all of the APIs a browser like Chrome already ships along with a Javascript engine. Maybe if we could break those all up into cross-platform shared libraries it could be done.
- Jtsummers 9y agoIt only has to support the DOM and web technologies if we decide it needs to be compatible with web technologies. That's not actually needed for a cross-platform toolkit like this. Regarding your link, it's for JavaScript -> Android Native. I found another project that was an Electron-compatible effort on top of Gecko. I haven't done much other searching for things based on the name, honestly Positron just popped into my head while reading the original article so I used it here.
- agentultra 9y agoI think in order to compete with Electron and the developers who use it you will need to support the web technologies and APIs it has.
- mcphage 9y ago> Is your code so tightly coupled between logic and display that you haven't reduced the common features to, essentially, libraries at this point? Separation of concerns is not a novel concept. No, I don't think many development companies have reduced the common features of codebases in 3 different languages to libraries yet.
- Jtsummers 9y agoDifferent platforms, not languages, per the posted article. Of course it's hard to combine multiple languages into a single library set, but if you're dealing with multiple languages you're often already dealing with multiple applications anyways and not combining them into one singular application. Or you've figured out FFI and can deal with the multiple language issue that way, in which case, wait, you have libraries in other languages!
- mcphage 9y ago> Different platforms, not languages The article mentions iOS development in Swift, Android development in Kotlin, and web development in Javascript. > Or you've figured out FFI and can deal with the multiple language issue that way, in which case, wait, you have libraries in other languages! As I said, not many development studios do that.
- Matthias247 9y ago> Hell, Elixir is popular, let's go with that. You'll get great concurrency from the start, and hot code loading, and a decent language Just a comment on that: Different use-cases have different solutions. While the Erlang/Elixir model might work great for servers we gained the knowledge that a singlethreaded event-driven model seems to work best for user interfaces, because they are very stateful things. Multithreaded user-interface technologies (which is what Erlang would be - even though you have actors instead of threads) have afaik never worked out. Javascript however embraces the singlethreaded event-driven model up to the point where no shared-memory parallelism is possible - which works well for user interfaces. You don't even can't have the error anymore that some callbacks runs from another thread than expected (which is a typical error in GUI implementations in Java and C#). And another remark: If the main goal of the change is to improve performance and resource usage, I also don't think Elixir would bring us forward. The erlang VM isn't a super-high performance environment, most likely electron/node/V8 is faster.