Request to make a working fork of the emmet-atom package canonical

[wideaperture]

The emmet-atom package currently available through the Pulsar package manager crashes whenever it's used. Github user @backit forked the package and created a fix. Would the Pulsar gods be so kind as to make the fork into the official version available through the app's package installer? Installing the fork manually is simple enough, but I'm using Pulsar with some students this semester who've never coded before and I'd like them to be able to install a working version of emmet-atom from the GUI package manager. They're still at the stage where pulling up the terminal and entering an npm command is going to give them math anxiety.

[savetheclocktower]

EDIT: The underlying issue with emmet has been fixed in the latest rolling release and should make it into the 1.130.0 regular release. You'll be able to return to using the official emmet package.

---

(I've had the official emmet package installed for years and it's never given me trouble — though I'm now hesitant to uninstall/reinstall and tempt the wrath of the gods.)

It's not unreasonable to want what you're asking for here, but it would require us to open Pandora's Box. I know it's unlikely, but suppose we change what emmet points to in our package registry… and then later the emmetio organization updates the emmet package and wants to publish it to our registry. What happens then?

That's assuming that everyone involved is acting in good faith, too; there are more cynical concerns around supply-chain attacks which I'm sure don't apply in this instance. But in the general case, if a user has installed a Pulsar package from one author, they've granted that specific author an implicit trust, and it's wrong for us to transfer that trust to another author without offering them a veto point.

Assuming the most likely outcome (emmet never gets fixed), here's the ideal outcome:

• If we confirm that the issue with emmet affects everyone and there's no simple workaround on the Pulsar side, we'll add a “broken” badge to emmet on the package repo and guide people toward other options.
• One of those other options could be the fork that @backit made. You were right to request that they publish their fork to the package registry. (Naming is always a pain in these situations, but I've tended to add -redux to the end, uncreative as it is.) It's been a week since you asked, so I think you'd be entitled to fork their fork and do it yourself — and then your package could be one of the other options toward which we guide people.
• pulsar-cooperative is one of our many half-started projects, and aims to be somewhat of a solution to this. It's designed to be the home for unmaintained packages that need only small fixes to keep working, but for which nobody wants to assume de facto maintainer status just because they made a one-line fix. It's almost certainly the best option in the long term, but until we set up a CI pipeline that can publish directly to the registry, it doesn't get you want you want (an installation that doesn't involve the terminal). But this is a reminder that we could be making this whole process easier by trying to get unstuck on pulsar-cooperative.

Anyway, for now I'd suggest forking @backit’s fork and publishing it to the registry on your own. If you're willing to do this, you can read the docs and give it a shot, following up in this thread or on Discord if you run into trouble. Be sure to update the repo URL and the package's name in the package.json. (The package name and the repo name don't have to agree, and in fact already do not in emmet’s case.)

[savetheclocktower]

OK, now I'm thinking

and there's no simple workaround on the Pulsar side

is not correct. The root issue is the assumption of the presence of a tabStopMarkers property on a SnippetExpansion instance, and while that's not been present for a while, we could certainly pretend it existed if it fixed this issue. Let me see if I can put together some lines to throw into an init.js to test a possible fix.

[savetheclocktower]

OK, @wideaperture, try this:

• disable other forks of emmet and install only the (currently broken) one from the registry.
• drop this code into your init.js and reload your window (or relaunch Pulsar) to see if that makes emmet work again.

It seems to work for me. (emmet has only been working for me because I remapped it to a non-Tab key, and this exception is thrown only when you invoke emmet:expand-abbreviation-with-tab.)

The actual fix is nowhere near as silly as the code in that Gist; we can probably get it into the next Pulsar release, and then everybody wins. (This is not a scalable strategy for package failures, but emmet is very popular, and the fix is very simple.)

[wideaperture]

Will give it a try tonight! Thank you so much.

[wideaperture]

Sorry I didn't get around to this yesterday evening. I saw that you subsequently merged the fix into the latest rolling release, though, so I downloaded that and it seems to be working for me! I greatly appreciate you doing this and thanks as well for moving so quickly on it. I can tell my students this morning to grab the latest version.

[savetheclocktower]

Honestly, if I'd read #1132 more closely a while back, I would've realized how quickly this could've been fixed and saved a lot of hassle!

[Raf-sns]

Hello,
I read your discussion but didn't understand everything. Has the bug been resolved?
What should I do to resolve the issue?
Can you provide clear instructions for an update if the issue has been resolved?

Thank you for your help and dedication. Kind regards, Raf

++ - I encounter this error when i tried your code for the init.js file:

Failed to load /home/raf/.pulsar/init.js
whenPackageActivates is not defined

[savetheclocktower]

I left a function definition out of that gist. How silly of me. I'll update it.

The bug has been resolved, but the fix is only present in rolling releases. If you want emmet to work right now, you can download the latest rolling release of Pulsar — or else wait until the next regular release, which will also contain the fix.

My gist is pretty hacky and I wouldn't rely on it at all, but if you don't want to switch to a rolling release and don't want to wait for a regular release, that's probably the third best option. Just remember to remove that code from your init.js once you update Pulsar.

[savetheclocktower]

For those reading this in the future: version 1.130.0 is (or will be) the first regular release to contain this fix. If you're at that version number or higher, the original emmet package should work just fine.

[Raf-sns]

Thank you very much, your code works wonderfully.
I'll wait for the next Pulsar update to install the final version.
All these little problems related to Pulsar that are piling up were starting to make me think about another code editor, but I definitely don't want to install any Mxxxxsoft software on my PC.
I have so many configurations on Atom/Pulsar that I can hardly live without it.
Thank you for your responsiveness and your time. Open source and collaboration help us move forward so much as programmers and humans. Bravo!
Sincerely,
Raf.