Maybe I'm just an outlier, but as a front end developer, one of the _last_ things I want to do is spend time hacking my tool chain. What gets me excited is delivering new features and better UX to my users. Even a minute spent worrying about my development environment is a minute taken away from what makes me happy. Truth be told, my main complaint with VSCode and atom isn't really the non-native performance of Electron, but all the effort required to tweak, configure, keep plugins updated, replace outdated plugins with the latest and greatest, etc. Seems to be a microcosm of the entire front end tool chain these days. (First it was npm, then yarn, then back to npm; jasmine, no jest, now karma; LESS, SASS, postCSS; Babel, no webpack, what about parcel; etc.)

Even though my current work is all Vue and Angular, with a backend that's little more than a REST facade for MongoDB, I still find Coda to be the best development environment available, and I continue to use it all day every day. Considering how strongly Coda was originally tied to the then-dominant LAMP stack, I think that's a pretty strong testament to the product. (Or perhaps just a testament to my stubbornness.)

I won't buy the next version of Coda sight unseen, but it's hard to imagine what Panic might do that would keep them separated from my money.

What are you talking about? What tweaking? You just install extensions and they’ll auto update. You only need to configure things when you don’t like the default behavior. I haven’t touched my VSCode configuration in ages.

The counter to your position: You optimize your dev environment once, and it pays dividends. Relative to the return the initial cost of optimization is negligible. Especially when with editors like vim my configuration file is stored in github, so I can literally replicate my precise editor's configuration, with all my personal tweaks and plugins, in 30 seconds, on any computer with internet on it.

You interpreted hackability = ability for you to spend your time hacking on it.

But I think they were going for hackability = other people will be more likely to develop plugins. The lower barrier to entry is basically one of the major upsides of Atom/VSCode: their plugins tend to have cutting edge support.

For example, iirc Atom was the first to have a good plugin for Elm lang (elm-jitsu). Meanwhile the Elm plugin for Jetbrains' IntelliJ came later as the result of the effort of two Microsoft employees spending billable time on the somewhat academic project.

A better example is that the creator of elmx ("JSX for Elm") was able to create an Atom plugin for their basically experimental library: https://github.com/pzavolinsky/elmx#atom-integration

I also think this differs from "hackability" when used with Vim/Emacs which does mean "you will be hacking together your .*rc file" which is definitely not how I'd characterize VSCode/Atom.

You sound like preparing to use VS Code is like preparing to get vim right for your use. VS Code is far easier.

"Newest and greatest" can be achieved so easily, not sure what you're complaining about.

The real strong point is it has strong community of plugins to fill your use case which Coda is missing.

Yeah. I really love Panic, but given the zillions of developer-hours poured into competing text editors, it's super hard to imagine a small team (even a talented one) competing.

There is room for innovation in the rest of the web dev workflow, which has gotten a bit weird since almost everything involves a compilation step now. Not sure how this could be achieved, but I guess that's why I'm not the one doing it!

Two very influential editors come to mind, both developed by a single person AFAIK: TextMate and Sublime Text.

It may be hard to sustain it in the long run, but they both changed the landscape at the time.

Definitely. TextMate was my primary editor at one point, and Sublime is still my editor of choice. =)

But, I think the bar has been raised since those days.

Sublime's been Jon's fulltime work for a decade, and has probably 20,000+ hours of work on the core editor, at least an order of magnitude more work put into its extensions, and certainly millions of hours of real-world use and bug reporting.

That's kind of the minimum bar that newcomers need to approach or exceed just to get into the game in 2019.

Unless they are bringing something totally different to the table. For example, Atom's Electron base has drawbacks but it brought a new level (or at least a new kind of) hackability to the game.

In Sublime and Textmate's case, they too brought something unique to the game. While lacking the raw power and configurability and pedigree of vim and emacs, they brought something of that mentality to the world of GUI text editors. And their minimal GUIs allowed them to be flexible and powerful in ways that full-blown GUI monstrosities like Eclipse could not be.

Sublime Text and TextMate were different developers. Sublime appeared during the extremely long (years and years) development period of TextMate 2.

Sure, I meant they were both developed by a one person only shop. Not that they were both developed by the same person.

Also, Sublime was Windows only at first IIRC, so I don’t think it was an answer to TextMate.

Sublime was pretty much "TextMate on Windows" when it was released. Then most TextMate users abandoned ship when it entered development hell. Then Sublime went cross-platform. Then Atom came and took most of the cake. Then VS Code came and took so much of the cake that there's barely anything left.

The Stack Overflow Developer Survey 2019 suggests the cake is more distributed than to you think.

Unless you mean the "blog post headline" cake, in that case you might be right.

Sometimes it seems the "blog post headline" cake is the only cake that matters.

I’m sure there is an opportunity here, I looked for a native editor for MacOS but the pickings were thin.

Yea, as an example I found two plugins just this week that have hugely upgraded my productivity:

1. JSX <> Object convert

2. CSS <> Object convert

I don’t use a ton of plugins, but there are about 12 that really make my life far easier. That’s probably the same for everyone, but a different set of 5-20 plugins.

That, and the fact that Intellisense/Typescript features are about all that matters in terms of future productivity gain, and I can’t imagine why they’d choose to make this. By the time they release it they’ll be that much further behind the status quo in VSCode, which is far ahead and moving fast. Best of luck to them!

I think the big difference between Panic's editor and VS Code will be that it's a native application, that it's not using Electron. In the past Panic has really set themselves apart by leveraging Mac OS, I don't believe they are much interested in developing cross-platform applications. That tighter integration with the platform may also provide advantages over tools like VS Code, where there would be a real cost to developing OS specific features.

For me as an Ops and automation, it being native rather than a web-frame / JavaScript app means a lot for latency, performance and OS integration.

Hopefully it also has good PostgreSQL support and plugins / module support for languages / DSLs such as puppet, SQL, ruby etc... and decent git / GitLab integration.

Oh nice! What kind of academic research?

See my website for more details: http://web.eecs.utk.edu/~azh/

I focus on the usability of dev tools, including features for navigation, refactoring, testing, and code reviewing. For example, a big part of my dissertation was Patchworks [1], which provides a never-ending carousel of code documents as an alternative to the traditional tab document-based editor.

Wow I love what you do there! I am myself crazy for development tools. I think I tried almost every editor available for MacOS and Linux. I hope that the future will bring us even more AI driven code intelligence and context based suggestions.