Show HN: Unforget, the note-taking app I always wanted: offline first, encrypted (unforget.computing-den.com)
Hi HN! I created Unforget out of years of frustration with Google Keep and the lack of alternative that met all my needs. I hope you find it useful too!
Features include:
- import from Google Keep
- offline first including search
- sync when online
- own your data and fully encrypted
- Desktop, mobile, web
- lightweight, progressive web app without Electron.js
- markdown support
- programmable with public APIs
- open source [1]
In addition, I'm thrilled to announce the opening of our software agency, Computing Den [2]. We specialize in helping businesses transition from legacy software, manual workflows, and Excel spreadsheets to modern, automated systems. Please get it touch to discuss how we can help you or if you wish to join our team.
144 comments
[ 3.0 ms ] story [ 192 ms ] thread2. Nit - on the demo page: Priavcy --> Privacy
3. Why does one of my browsers raise the error, "Your browser does not support service workers. Please use another browser."? Can this be worked around, or is the use of service workers essential to the distinguishing characteristics of the app? Genuinely curious.
I'm curious which browser are you using?
I don't know much about service workers. But, after some searching I found out they apparently aren't supported in private / incognito mode.
I do 95%+ of my browsing in incognito mode, so this makes a lot of sense.
After refreshing, it properly displayed the JS code for the SW.
Tor Browser
> Well without the service worker it wouldn't work offline any more and that's one of the key features.
Thanks, that definitely seems worthwhile then!
* Is this a bunch of titled markdown docs organized (conceptually) into folders/hierarchically?
* Is this a bunch of untitled/title optional “cards” organized by tag?
* Is this a long, markdown document, which you append to?
These are similar to different mainstream and less so note taking systems, and would appreciate understanding what workflow you’ve designed and optimized around.
Saying you started as a Google Keep user is helpful, buy I’ve only used other systems (Homebrew textfiles, Simplenote, obsidian, etc), and have some concepts around what Evernote and OneNote are like, so giving a couple more signposts on usage would be helpful.
To answer your questions quickly, I usually keep very small notes just a few lines or todo checkboxes. One note per idea. But sometimes an idea grows over many days and that note gets much larger. There's no limit to how large a note can get.
You can always set the title of the note using # which is standard markdown, or leave a blank line after the first line and it automatically becomes the title.
There's no concept of tag per se but you can write #someTag and then literally search for #someTag. The search feature is just substring search over all the notes (no stemming or anything fancy).
* You have a big chronologically ordered list of notes
* By default, all notes are in view
* You can make a title (instructions in demo), but significance of title is only internal to note (not for ordering/management)
* Big list of notes is union of two disjoint sets: pinned and unpinned
* Can view either all notes, pinned notes, or un-pinned notes
But not sure if I have that right. FWIW appreciate some of the design decisions I’m seeing, just haven’t had time to poke around to understand.
In the beginning, I intended to add dragging of notes to change the order and organizing notes in folders. But, honestly, after using it for a while I realized that I don't really miss those features at all even though I have thousands of notes. As long as the search is fast I can always find what I need quickly.
Again, thank you. I'll add this info to the demo note.
FWIW agree with your call on not building too many organisational tools in
Thanks again for sharing; look forward to trying it out
I also build a web-tech-based note-taking app, and honestly I resisted wrapping the thing in Electron for a long time for the same reasons HNers love to hate on Electron wrappers, but honestly people like having an actual "app" (even if it is Electron) way, way more than they enjoy using a web app, in our experience.
You can still access our app via web only, but almost nobody does because they like the apps much better.
> but honestly people like having an actual "app" (even if it is Electron)
Do you mean because of the tighter platform integration and more features? or the fact that it is an icon on your home screen? if you mean the latter then a PWA is actually quite nice. Some non-techie people I showed this to were quite impressed and excited to find out that they could just add a website to their home screen that acts like an actual app including working offline.
This reminded me of Joplin. What's the direction you're looking to take this in?
For example, the workflow didn't seem very fast to me. For example, open the app, then at most after one tap you should start typing immediately, enter multiple todo items without going to menus or pressing any other buttons. On desktop, it should all be keyboard driven anyways. In other words, I want the UI to get out of the way. I think this is something that many note taking apps miss.
As for the direction, I'd like to make it easier to migrate from other apps. There's already a Google Keep importer but there should be many more.
Also, with the public APIs, I'd love to see some Emacs, vscode, vim clients/integrations :)
Other than that, it pretty much covers all my needs at this point.
To install it, in safari go to share -> add to home screen. That's all!
Webapps are the thing.
If Apple lets them be - or is forced to.
Is another form of organizing essential for you?
Tags could cover most of my need for taxonomy/organization. (I'm hesitant to say 'all'.)
FYI I noticed a typo, "Github falvored markdown".
[0] I try for 'inbox zero' but am okay with leaving things in my inbox for a few days if I'm expecting to need them soon. Sometimes I flag for followup or color-categorize.
-- Allow regular expression search. -- Click outside the note to save. (Autosave every N seconds would be nice as well.) -- Customizable CSS. (Yes, I can do this with a local build.) -- Collapse/expand (sub)headings.
I'll give it more of a test drive over the next few days.
Auto save is already done, nothing you type is every lost. It's saved immediately (and synced every few seconds).
There is a collapse/expand feature but not for subheadings. If you put --- at the beginning of a line followed by a blank line it'll collapse everything after that, as shown in the demo note.
I did see the '---' collapsible and that helps but only if you have one section you want to collapse. I'm perhaps more visually-oriented than most, and being able to hide extraneous material helps me focus.
- level 1: collapse all H1 sections
- level 2: collapse all H2 sections
- ...
Basically like org-mode's cycle feature with the tab key. And then clicking on a headline will expand it.
Why did you choose to build this with a webdev stack?
(Not a criticism! Use what works for you! I'm just interested in why people choose the tools they do.)
Also, even if we exclude the web, you still have to deal with app stores, code signing, separate builds for each platform and whatever else you'd have to do to distribute your app everywhere (not my area of expertise). But with a PWA, you make a website that uses Service Worker and you're done.
With all downsides of the web stack I still would rather go with it than QT.
[0] That is to say, before the macOS rebrand, so this was a number of years ago.
That's important.
* It's (probably) sandboxed
* It's easy to distribute
* It's easier to make an app with a decent (but non-native) UI, compared to other cross-platform toolkits
Typescript is about the best mainstream (not Haskell) programming language I've ever worked with, modern web frameworks (React, Svelte and others) use functional programming concepts in the right places without requiring knowledge of type theory, it's easily debuggable at runtime and modern css/html is actually a pretty good way to do layout of complex user interfaces.
Web dev just feels insane to me. The amount of logistics and infrastructure and tooling you need[0] is beyond the pale.
[0] "need" is maybe strong, but in an enterprise setting like I was that's what you have
Guess that complexity is the same as number of spinners.
[1] https://connect.mozilla.org/t5/ideas/bring-back-pwa-progress...
Why do I need to sign up? When I "sync" how do I know my notes are private? Where are they 'synced' and who has access to the keys?
I wish there was a 'how Unforget works' section
I'm not web tech savvy, but I value my privacy (especially with notes) and I need a note taking app with permanence. With this landing page I have no idea how it works, where my notes get saved, how do I access my notes if https://unforget.computing-den.com/ goes down. how do I self host?
maybe I just didnt spend enough time or navigate to the right section of the app shrugs
edit: I was going to delete this comment because I am not sure it adds anything meaningful. It was mostly a rant. But I want to explain that I'm coming from a place where I've also been looking for a note taking tool for a long time and this is very close to what I'm looking for.
Show HN's are for people to show their work and get constructive critique/interesting conversation. Take a look AT https://news.ycombinator.com/showhn.html
It was constructive critique. Terse, but not only did I point out the things that are unclear to me I suggested a fix.
The edit was to soften the effect of being so blunt. Its a technique of being harsh then giving the reader some reason to attribute their negative emotions to. Its a way of side stepping the ‘defensive reflex’
So my friend I would appreciate you take as much time to read as write
[1] https://github.com/computing-den/unforget
You can self host of course in which case you only sign up to your own server. Instructions for self hosting can be found on the github page [2].
As for verifying the privacy of your data, this is a little more complicated. First, since it's a web app, you can open the dev console of your browser and in the network tab you'll see all the data that gets passes around. You will see that the content of your notes and passwords are all "gibberish" because they are encrypted or hashed. Of course that doesn't necessarily mean they are encrypted or encrypted well. To verify that, you really have to take a look at the code at least the client examples [3] which is provided in both typescript and python. They will show you how encryption and decryption work which allows you to verify that the "gibberish" data you see in the network tab of the website really is properly encrypted notes.
I hope that addresses your concerns. Let me know if you have any more questions.
[1] https://en.wikipedia.org/wiki/NAT_traversal
[2] https://github.com/computing-den/unforget
[3] https://github.com/computing-den/unforget/tree/master/exampl...
Also, thanks for the markdown support. I'm one of those weird people that will ask if you'd be willing to support org mode as well. And I'm kinda surprised you don't if you use it yourself! I'd be very likely to use this if I could simply point it to an org file and have it displayed nicely on the web. I often take short notes that turn into long notes or vice-versa.
A fair response to your question would be, "submit a pull request" (thanks for being open source)... but I've got far too much science to do at the moment. So, regardless, thanks for building something cool!
Since the APIs are public and documented, I would love to see a client inside Emacs as well. Then we can browse the notes in actual org-mode on desktop.
I can't say I have any plans for it yet since it covers all my needs as is. But, if I find some spare time, I'd love to make it happen.
There's an appeal to it, basically using the web purely as distribution, and the browser as the runtime. But I'm concerned with the fiddly-ness of local storage and such like that. The "out of site, out of mind" nature of it. The data not being in a "~/.app/app.dat" file, etc. The idea of it potentially just up and vanishing with a browser update. And, heck, just the complexity of dealing with the schema versioning in the native web model.
As well as the portability of data (say when you copy over to a new computer).
The idea of simple web distribution is compelling. Now you need nothing more than a github account, and off you go. No server, no nothing.
But I still feel (perhaps ignorantly) that the data situation is still on shaky ground. (Discounting the whole lack of something like SQLite, etc.)
And, the idea of bundling something like SQLite as a webassembly blob just makes me itch.
Just curious how that's worked out.
(Although, I don't have experience with SQLite, so that file might be as volatile as `localStorage`.)
[0] https://www.npmjs.com/package/better-sqlite3
[1] https://github.com/computing-den/unforget/blob/master/src/se...
Also, wanted to mention that Unforget doesn't use browser's localStorage. LocalStorage has a low limit and the browser can nuke it when it wants. Instead we use IndexedDB which has much higher limits and persistent.
Having said that, indexeddb has been very reliable. There are some annoying limitations though. For example, if you want to create an index on a store using multiple keys, you can't specify which key should be ascending or descending.
If you add the PWA to your home page, i.e. install it, then the browser must not clear indexeddb on an update or something. And as far as I can tell, browsers do respect that.
My overall experience building PWA has been mixed. I think it really depends on what app you're making. For a note taking app, where the data can be backed up to the cloud encrypted and synced, I really think a PWA is the best option. But if you want tighter integration with the platform e.g. accessing the file system, then I would stay away from PWA.
Another issue is ios safari which is missing a lot of features for PWAs and I think unlikely to support them in the future. E.g. background sync, offline push notifications, etc.
Are you considering PWA for any particular app?
So, I was considering a PWA. I don't need tight local integration per se. I do have a conceptual problem not being able to save a file or read a file, even if it's just chosen by the user. I THINK I can do that. I THINK you can pop a file chooser to read a file, but not necessarily just open any file you want. If you can read/write a file as directed by a user, that would be fine. I can see the rest being stored in the IndexDB. And that gives the option of an export/backup if it becomes necessary for peace of mind.
Then there's the whole x-browser game, I'm not really looking at mobile, just desktop, but I'd like to support FF/Chrome/Safari/Edge.
It's not rocket science stuff by any means. But I don't want a user to call up and talking about how their data suddenly vanished because they went to get the latest, and I did something innocent/ignorant (or not), or THEY did something (deleted their PWA icon, copied it, moved it, etc.), or the browser did something.
I have no experience with it, so for now, I'm not comfortable because I don't have my "hands" on the data.
If I can export it, then the comfort level goes way up. Inform the user to back up "just in case", especially early on when we're all still bumping in the dark.
To be clear, I don't want a server side, I have no interest in standing up a service component. I have less interest in maintaining a service component. I'd like the code to be local and "standalone" as practical.
But I also want something a touch more than "git clone app; cd app; ./build".
I believe there are other libraries/frameworks on github with a similar approach you might want to have a look at.
[1] https://github.com/webview/webview
https://tauri.app/
This creates Unforget.app in ~/Applications which appears to work. I've also successfully used this with Tana in the past. Did you run into trouble with that or have you not tested it?
Another similarity half-baked plan is for a PWA to regularly post (encrypted?) data to a service I'd host that simply forwards the data to the user's choice of cloud storage. It would require initial setup and probably regularly refreshing token, but should be workable.
[1] https://actualbudget.org/
[2] https://www.youtube.com/watch?v=2dh_gtndayY
[3] https://www.localfirst.fm/7
Anyone writing applications in the browser should be very cautious to note that many browsers will jettison some types of their website storage if disk space is low.
This is frustrating at the best of times, but if there's no server-side backup then frustration would be an understatement.
I am not sure if this caveat applies to localStorage but from experience it does to IndexedDB — seemingly even in Electron applications (??).
oh fun! I haven't run into that. But if you're using electron or any platform's local webkit wrapper, then it makes a lot more sense to forgo the browser's storage and directly use sqlite + server-side backup if possible.
It is possible to do this now in chromium based browsers. Refer https://developer.chrome.com/docs/capabilities/web-apis/file...
As far as I know, it’s using some caching library at the HTTP level (as opposed to handling its own persistence at the application layer), and in falls apart every time I have a poor connection, remaining broken (even after fully going offline) until I’m back online.
It works pretty well with the only downside being the user needs to re-select the file/directory to use each time they use the app and there is no way to force the file picker to start in a specific directory other than the "well known directories".
https://developer.chrome.com/blog/persistent-permissions-for...
BTW anytime a notes app comes up I feel the need to let people know that there is an option to bulk download your apple notes at privacy.apple.com
- everything end-to-end encrypted by default
- public APIs, create your own client (still with e2e encryption)
- self host
- markdown support
- import from google keep
- open source
Ah yes, you only need... an entire browser to access it, so much better!
Obsidian is also extremely simple and only offers options to expand its features, without ever mandating them. I'll also add that simple is not always better, especially (like this "app") when it's simplistic rather than simple.
It also, you know, works when you don't have internet?
electron is basically a browser wrapper in this case so yeah this is just skipping an unnecessary dependency.
not everybody uses obsidian. everybody has a browser.
No, it's a very stripped down browser.
Firefox with 0 browser extensions and only opened on that note taking website consumes 450MB of RAM, while my Obsidian installation only consumes 136MB.
Even just looking at that specific tab (on Chrome because Firefox doesn't show it), it consumes 86MB of RAM on the demo. I'll happily spare 50 extra MB for all the features of Obsidian.
Wrote a post of different ways you can sync a Obsidian vault/project: https://bryanhogan.com/blog/how-to-sync-obsidian
"Techinal details"
fyi
if (a.modification_time > b.modification_time) ...
that's all. Like some other features I originally planned to implement CRDT but then realized I don't miss it. Since this is not a collaborative editor with multiple users editing the same note, CRDT is not that useful. I think the only situation in which you'd want CRDT for a private note taking app is if you're editing the same note from multiple devices while all are offline.
[1] https://github.com/computing-den/unforget?tab=readme-ov-file...