Questions about daily-driving Zedless

[digitalextremist]

Moving over from: https://github.com/zedless-editor/zedless/issues/39#issuecomment-3265826041

Curious to continue hearing about your case @max-privatevoid.

`NODEGATE` caused me to rethink `Zedless` as I ran into `node.js` using more resources than `Zed` itself.

I am primarily working across some pretty heavy Typescript projects, some with Vue and some without, some front-end, some back-end, a lot more that are neither; as well as a large library of bash scripts and nginx configurations. There is more but that is a gist. I do represent the 'javascript' use case, but plan to head back to Crystal ( or Ruby ) again mid-term. My design-side is language agnostic and I am using Typescript with a loose grip.

#23 was refreshing; I am not in the node.js contingent, though I philosophically align somewhat ( taking the trademark back from Oracle, etc. ... but it feels futile/short-sighted ) ... but as mentioned Bun is VC-code also, whether run by the IDE or running my code... tricky, tricky. Everywhere is an agenda to keep in my sights. Meaning, take up context space, degrading quality of life.

As soon as I shifted my wife over to Zed I really started to 'feel' the issues that I had just unconsciously began ignoring in the interest of expediency. Once I got her on that managing her library of Markdown files, on two of her machines, it started to be weird to me that setting up Zed for normal use without invasion was like preparing to go on a deep-space mission millions of light-years from Earth. A lot stops working silently; stuff behaves differently on almost identical machines. Insanity is a common feeling.

Then as of the beginning of NODEGATE it started to dawn on me that in the next 5-10 years I might hate life so bad it will become unsustainable, but I might not have noticed and would have just slogged through it.

I believed-in and saw */OSS there, but I see Mo Money Mo Problems ... it is getting to be extremely annoying to deal with. My own work feels secondary to the tool, and do not feel like there is any real art or craft left, only better or more or different ... which is not the same thing. I don't have VCs to impress. I am happily married and doing actual work in the real world. I do not like feeling like there are paid friends in chat on one side, and the salty-spectrum on the other, with corporate all over the place and no real */OSS soul building. There is a particular kind of stress telegraphed by making $XXm promises and needing to follow-through. I am not trying to compete with the IDE world, nor unicorns. There is emotional burn-out trying to keep up with that to do my own work. I am in an IDE 12-18 hours per day right now, which means Zed ... lately have been wishing for vim which is a bad sign.

It felt weird/wrong to speak against Zed since it is the best IDE out there, but... I want my life back. [ And now I was silently banned without explanation. ] I feel the human cost more than I ever did in VS Code or Sublime Text or TextMate ... when I open up vim I have this feeling of... 🧡 🚀 and I guess I miss that >90% of the day.

Now I am wishing to get off the trip and just have an IDE, rather than be on an ongoing adventure.

IF, and this is a big IF, the best of Zed comes over:

• Do you use any LLMs? I cannot recall if that was something that got tarred and feathered or burned at the stake here.
  • Would personally continue to use Ollama but rip out everything else.
  • Would support ACP once it supports Ollama Ollama feature-parity for ACP agentclientprotocol/agent-client-protocol#51
• How often do you plan to pull in improvements? It seems like there is one pre-release from around 5 Zed lifetimes ago.
  • Do you plan to do things like track Quarterly changes and curate changes versus just pull in the fire-hydrant of update?

I would work with you on Zedless ( no matter what it be called #38 ) and go-long on this, IF as above.

I do not need improvements every 72 hours or so, and would be very happy with Quarterly or so if it worked 99% the rest of the time. Right now it feels like any time something changes in several worlds I want nothing to do with, there is a ripple-effect in Zed ... it has very noisy neighbors.

If seems we agree Zed is best, but it has no full-time conscience, and makes dubious decisions at high speed. I see this as keeping VC-code honest, or maybe just running off. I am preparing to jump off the train I actually have to. The pain is getting to be a lot.

I no longer believe Zed Industries, Inc. has any interest in the roots of why */OSS exists. And I do not believe they are free and/or able to care.

[digitalextremist]

First time I've seen one of these pulled:

zed-industries/zed#33165 (comment)

Looks like I harshed up your PR pretty hard there @nakamorichi 😬

[ and then I was silently kicked / banned a few minutes later from the Zed discord; wow... that was a fast cultural shift... ]

[digitalextremist]

Refactored my post after being banned without explanation. Seeing the cold and ruthless aspect of Zed was illuminating.

Am looking at neovim and this fork, hoping for this fork, but it seems like it will be a tall order to track changes. It seems like there is going to be a point when Zed becomes so much about chasing market-share and pleasing/placating the vibe-coder contingent it is going to become trash. Perhaps there is a point to just freeze it at.

[max-privatevoid]

My stance regarding LLMs: As I don't want Zedless to be tied to non-selfhostable cloud services, I will not allow any integrations specific to cloud-only LLM services. I am completely fine with local AI and my plan is to generically support OpenAI-compatible API providers, as that has become the de-facto standard for most FOSS inference engines. I will also keep the Ollama integration because it was already there. I don't care if anyone uses cloud-hosted LLMs through an OpenAI-compatible API, but this project will not build and/or support special client integrations for non-standard APIs.

Pulling in upstream changes: I do this pretty arbitrarily, usually whenever there are interesting/worthwhile changes in Zed and I have time to figure out the merge conflicts. I would not be opposed to the idea of becoming a "harder" fork and only cherry-picking a particular selection of commits or PRs rather than merging everything.

Alternatively, I also have some scripts that can delete certain files, module imports, function definitions, function parametes, enum variants, etc. using a combination of globs, regexes and tree-sitter AST walking. Those may be useful in steering the project more towards the way ungoogled-chromium operates, automatically applying a bunch of modifications on top of the Chromium source tree. Zed isn't quite as huge as Chromium, so it might be possible to maintain a hard fork with fairly little manpower, but if that isn't the case, the automation route would allow us to rely on Zed for maintenance (at least until they have burnt up all their money).

[digitalextremist]

I appreciate your time responding on this @max-privatevoid. I am trying to understand your principles and I hear a BDFL stride developing, fork notwithstanding. That is really important, as long as you can be trusted and stick to your principles. We will see.

I hear a hard-no on ACP ( non-standard API ) and realized I had been caught by light-fantastic there. I went to close that post, and noticed I have been blocked from any actions on GitHub over in the Zed world, not just discord now, except for posting more great bug reports. Made me realize why the term 'rugpull' was immediately mentioned in the README.md here. And I ( now ) agree investing in non-standard APIs is unwise. How you framed your position on what to support seemed to be well-thought. Seems like a lot of effort has been put into separating programmers from their ideas and work, which I had missed, under a spell.

It seems you have thought a lot about key chores and details, have good comparables ( i.e. Chromium ) and a healthy dose of audacity mixed with real work. That seems fair to consider its own contribution at the level of an original project, though it is a fork... like a refinery vs. drilling vs. shipping raw materials in, and shipping produced materials out. There seems to be a place for Zedless. A lot of the IDE itself is about standardizing things that are ambient; making common sense easier to start from.

As I embraced my ongoing dark night of the soul, I started to look deeply into neovim during this. I found myself missing key pieces introduced by Zed which now I cannot say are unique to Zed at all, but more like conventions pulled out of the air, harvesting the ambient nature of being a programmer. I am looking at Zedless through that lens now; as a refinery, in my analogy above. It seems like this reasoning bears out because much of the UI is a straight up copy of Tailwind approaches, as was publicly stated, and much of the banter between figures out there shows that really there is a beach ball of aesthetic and sensibilities being codified in Zed ... All this leads me to my lasting question, and some possible ways to contribute:

• What really is Zed? What is just how programmers wish their tools behaved, which is more of a by-product?
• How can we prioritize what even matters within that, and not get tricked into the fancy, fly-by-night, the disposable?
• What will the cultural dynamic be? A BDFL structure? Is there interest in drawing programmers who care and don't bicker?
• How can we have a firm foundation for a decade and not retreat to old IDEs and not get slowed by the light fantastic?

It seems like the spirit of Zed is now about capitalizing on a bubble, a unicorn of unicorns given the >billions in valuations for IDEs. I understand the allure. But all that seems like unabashedly harvesting the art of programming, and then muzzling the offended. I put this all out there extremely standoffishly, looking to find a basis for trust, otherwise looking to just move on and replace whatever value was added, seeing that Zed is unfortunately pretty emblematic.

Seeing I was so enthusiastic about Zed I am reassessing my investment in software/mission I did not write, within my duties.

I would like to shift over to Zedless and give that a try in my instance; in the process probably help build up some of the missing pieces outside Rust and pinpoint trouble spots. And ideally help sort through what matters, what is even going on in IDEs, etc. Seems like a lot to ask one person to discern that, and yet it cannot be too wide a group and descend into a swirling madness.

So far I admire your recognizing early the need to fork, and the work involved; I hope there can be some stability and stride. Like I said, a lot of what makes Zed great is nothing to do with Zed at all, and it is going to be very difficult to point to its unique value. Especially as so much goes into the 'AI' code approach, it does seem there is a lot being recycled, spread under different names.

Also, background concern: Is GitHub itself the safest/best place for Zedless long-term, given the landscape?

---

By the way: My goal here is to have one solid decade of loving an IDE. I am devoted to helping it mature more than just evolve. I see a lot of energy lost orienting juniors and intermediates ( who I insist continue to exist, and be painstakingly invested into ) fussing with fads and 'tools' which end up being fidget spinners. There is this "to each his own" meme which is based on the natural world where that worked, but it does not work now. Learning to piss standing up for example, was great to find out really fast and early, rather than have to put that together over time by labor. How many would figure that out? Honestly.

Being able to teach a particular starting minimum setup, including IDE, seems important now, not just leaving new coders to their own devices, and not just having an "if it ain't broke, don't fix it" mentality either. Having mentored and led a few generations as well as argued for the entire time with salty dogs, while playing whack-a-mole with flashes in the pan... it seems realistic to expect one solid decade of pure awesome, in what gets done with the IDE, not just fussing with environments. Once there is ninja-skill, perhaps an individual will have a perspective to contribute. Until then, common sense ought be available, even ambient.

[digitalextremist]

Started building and going to attempt to shift over; seeing >1.5gb usage routinely on Zed and my environment keeps being stomped mid-session. Reliability is going down fast. Will see how I can contribute to this fork without getting into Rust and without depending too heavily on this fork. Seems like my goal is probably too much to expect without putting direct effort into it myself in some way. Will start by running the fork, and leave all this on the long-term horizon for consideration vs. answers under pressure.

---

Update: Up and running! Seems we are at 0.198.0 (debug) equivalence right now.

Scary/cool to see that my settings.json is coming over, and themes, etc.

Seems like full-separation is necessary but without that yet, moving over seems painless enough. About to head into a live session.

[digitalextremist]

Initial feedback on Zedless ... I am so much happier ... ~10x happier. I did not expect that. I am in the honeymoon period, I know.

I am dealing with LSPs #42 and wrestling with linters; but I can use CLI tools in the meantime. At least formatting is there, which is more mission critical and allowed daily-driving. This is night-and-day better; game-changer awesome.

I had gotten used to >1.5gb RAM usage, with consistently mounting lag, like that was normal; restarting the application ( major context switch with 5-10 instances open and set up "just so" ) every 20-90 minutes sometimes, or at least 2-4 times per day.

Thank you @max-privatevoid for spearheading this!

[digitalextremist]

Seems to be answered further in https://github.com/zedless-editor/zedless/issues/86 and after a significant amount of time using Zedless now, I am on the other side: still needing to setup LSP properly ( https://github.com/zedless-editor/zedless/discussions/42 ) and desperately needing updates to track Zed more realistically and sustainably