Comment by inglor
6 years ago
I mean, macOS already has built in automation (with Automator) that you can write in AppleScript or JavaScript JXA. https://github.com/JXA-Cookbook/JXA-Cookbook is not very updated but its wiki contains quite a bit of useful recipes.
Why would I prefer this over the built in option?
Hammerspoon is suited for tasks that keep working in the background. Plus it has access to stuff like arbitrary drawing on the screen, which AppleScript doesn't provide. The most used case is custom bindings for (almost) arbitrary hotkeys.
Automator/AppleScript are sufficient when you only need their standard hook points―services and such.
Btw, JXA is a big disappointment: it maps Objective-C/AppleScript's calls and especially structures very awkwardly instead of providing more JS-native experience. And afaik its official documentation is lacking.
AppleScript is awful to write and JXA is impossible to find documentation for.
One trick to finding JXA documentation is simply to look at the AppleScript documentation and translate the code to JavaScript syntax, or, depending on what you've doing, look at the Objective-C Cocoa documentation and translate the syntax.
For example, to obtain the current date and one day from now, in Objective-C you might write:
The equivalent in JavaScript is simply
> look at the AppleScript documentation and translate the code to JavaScript syntax
Yeah, I can’t do this. It’s non-obvious to me how it translates until I actually stumble upon the correct thing.
8 replies →
That's a strategy to guess code but it doesn't yield documentation.
And there's Fennel - an Lisp that compiles to Lua which makes writing Spoons (Hammerspoon plugins) pure joy.
https://fennel-lang.org/
If you like Fennel, you're gonna like Spacehammer - Spacemacs inspired Hammerspoon config written in Fennel. https://github.com/agzam/spacehammer
1 reply →
"...impossible to find documentation for"
Exactly. It's been years since first released and nothing is really out there to tell you how to use it. I wish they would finish it up as it is much better that AppleScript.
That will never happen. JXA is junk, same as Scripting Bridge before it; and they never shit to fix SB even when they had years to do it. There’s a reason Apple finally disbanded the Mac Automation team and fired the PM responsible. JXA should’ve saved Mac Automation, but it killed it instead.
Worst part is: it should never have come to this.
(For those that don’t know, I am the world expert at building production-ready Apple event bridges, with half-a-dozen under my belt. Outside of myself, Dave Winer’s Userland, and the original AppleScript creators, no-one else on the planet has ever managed this.)
Immediately after JXA was surprised-announced at WWDC14, I dropped everything else, reached out to Sal Soghoian (Mac Automation PM), and crash-coursed six weeks of testing and critiquing JXA, up to and including writing a nearly-complete “JavaScriptOSA” reference implementation for the JXA devs to learn from, or just outright steal.
Even rushed and unfinished, JavaScriptOSA showed what JXA should’ve been. And they had double the developers, all the AppleScript source and internal Apple documentation, and a year. You can d/l it here:
https://sourceforge.net/projects/appscript/files/JavaScriptO...
While the Xcode project is far too old to build now and some of that code is a lashup of which I’m not proud, the prebuilt component may still work in Script Editor (though I’ve not tested it in years), as should the standalone “demo” app I bundled with it.
Bear in mind, my JavaScriptOSA implementation[1] was based on my previous appscript work [2], which had alrady proven itself a genuine AppleScript replacement years earlier. Apple even considered including Python and Ruby appscript in Mac OS X Leopard, so it’s not like my code wasn’t already known about, read, or used by folks within Apple. But the Mac Automation team still thought they knew better than a production-proven solution that had at its height a couple thousand extremely happy users, whereas their own previous attempt had only proven an embarrassing failure.
I didn’t get so much as a thanks-but-no-thanks; 200+ work hours straight down the crapper. And JXA shipped half-baked and broken, and immediately abandoned to sink without trace. I scrapped my plan to write the book on Automating your Mac with JavaScript, because you can’t write really good books about really lame broken technology, at least not without losing your mind. (I should know: I lead-authored one of the last AppleScript books.) Apple’s last, best chance to turn Mac Automation around, pissed up the wall and forgotten just as fast.
Honestly, Apple’s real mistake was not firing that preening incompetent, Sal Soghoian, years sooner, while there was still something worth saving.
..
Okay, so with hindsight, JXA was never likely to succeed, even if they had done its implementation and marketing right. As an OSA language component, it lives in its own little closed world, completely separate to Node.js and its vast npm ecosystem. Even in 2014 Node was already killing it in the JS-on-server-and-desktop wars, but instead of “Embrace and Extend” JXA chose “Not Invented Here” instead.
Worse, JXA is built on OSA which is a ComponentManager technology; a hideously ancient, inherently insecure, and long-deprecated in-process plugin architecture. Thus the JXA implementation was already obsolete even before it shipped. A straight port of node and npm CLI tools onto JavaScriptCore with a V8 API compatibility layer for C-based extensions would have been the right choice. But the Mac Automation team didn’t see, because like their technologies they exist in a closed little world completely isolated from everything and everyone else.
One More Thing: Around the same time as Sal was getting his ass ejected from Apple, I wrote one more Apple event bridge, nodeautomation, just to prove my point that creating a competent AppleScript alternative for a massively popular, developer-friendly language is both quickly and easily achievable, so Sal’s team had absolutely no excuse for fucking it up twice.[3] Although it looks like recent macOS/Node updates have broken the build (again), so you’d need to run it on an older platform if you wanted to play with it.
I can still vouch for Python3-appscript, as that’s what I use myself, but I no longer provide support for any of them because as much as I love Apple event automation and the incredible solutions it makes easy to build, even I accept it has been utterly buggered to death now, and the only question remaining is how much longer till Apple finally put it out our misery for good.
..
That said, if any Hammerspoon fans wish to cross my palm with silver, I’ve previously offered to do a Lua-Apple event bridge for it, and would be happy to do so even now. (Misery loves company.:)
--
[1] No relation to Late Night Software’s original JavaScriptOSA, which like JXA was also riddled with flaws and never went anywhere.
[2] http://appscript.sourceforge.net/
[3] Or thrice, if you count the time after the JXA fiasco that I offered to give Sal my SwiftAutomation bridge [4] outright, just as Swift was starting its meteoric rise, and he threw it back in my face. I’ve dealt with some stupid PMs over the years, but Jebus that one was dumber’n all those other stumps put together. What a waste.
[4] https://hhas.bitbucket.io/
2 replies →
Applescript seems to have been a mistake in that it is too irregular and programmers don't learn and use it. I researched the most highly thought of books on Applescript and even they were emphasizing how hard it was to know thoroughly.
Key insight: Apple event IPC is not OOP, it is RPC + simple first-class queries. It is far more closely related to SQL and relational databases than to DOM.
Unfortunately, programmers look at AppleScript syntax, which is superficially OO-like, and assume it is OOP. Which causes huge confusion and frustration when it behaves in completely non-OO ways.
The confusion is confounded by AppleScript’s fondness for “magical” behaviors—including overloading native language operators and automatically dispatching remote calls (aka “implicit gets”) without obvious rhyme or reason—plus enough syntactic sugar to rot anyone’s brain to mush.
Ironically, non-programmers don’t struggle nearly so much because they are operating on simple ignorance rather than outright misunderstanding, so they accept what it’s doing on face value, rather than trying to map everything they see onto completely the wrong mental model and getting utterly honked that prediction and observation refuse to match up.
In fact there are hard rules behind all of AS’s behavior, but they are utterly opaque and never adequately explained anywhere, so are largely indistinguishable from “What the F* is Going On?”. Hell, I rewrote most of the early chapters for the last edition of Apress’s AppleScript book, and even I bottled it as “too hard to explain to anyone else”.
..
Things might’ve turned out differently had AppleScript’s original designers not quit in disgust at Apple [mis]management shortly after AppleScript 1.1 went out, but when they walked out much of the expert insight and knowledge went with them, not to mention the ability to fix it once users’ feedback was pointing all its flaws out. But “coulda, woulda, shoulda” is the Mac Automation story throughout (my own failed contributions included).
Dr William Cook’s paper on the early history of AppleScript and its design and ambitions is well worth a read:
http://www.cs.utexas.edu/~wcook/Drafts/2006/ashopl.pdf
Thanks for the link! Best explanation of Apple Events I’ve ever seen, plus interesting to read a first-hand account of AppleScript’s history.
1 reply →