Mine is online (rendered client-side) so I can access it from anywhere. https://start.oinam.com
Just in case, anyone wants to look the source is at https://github.com/oinam/start
The source is at https://github.com/wenerme/wener
I found it's very useful, especially the license and basic tech stack & description. One day I will grab the meta and display along with the link.
1. 20+ years of OS support for double-click to open
2. It already supports an embedded title.
3. It's basically just INI/TOML, so you can add custom fields that get ignored by parsers. Want to embed a screenshot? Just append Screenshot=B64(data). Want to add tags? Same deal.
You can also organise them as files in...folders.
That’s a dealbreaker for me. I like something that I can use grep and other Unix utilities on. There’s a reason that the tabular style formats of /etc/passwd, crontab and inetd exist.
> You can also organise them as files in...folders.
+1 on this concept, though. For some reason folks forget that the filesystem is a pretty decent hierarchical database!
Should be pretty easy to grep through. It's an absolutely minimal format
<html><head><meta http-equiv="refresh" content="0; URL='https://www.someurl.com' " /></head><body></body></html>
upon opening these htmls , the browser just redirects to the actual url.
To me they're completely different purposes. Tabs are short term - for whatever you're working on right now. It's like a short-term cache, a worktable to keep everything in your head right now easily accessible.
Bookmarks are long term, like a filing cabinet. You may not be using it today, but you want to hold onto it for later.
If I had to use either one for the other purpose, it would suck.
At the very least, I personally need tags. Tree hierarchies are too limiting, but tags allow adding free-form metadata which can be used to quickly find anything. I often don't remember the actual URL or title of a web site, but I'm looking for a general category, and with tags I'm usually able to find what I'm looking for. This does require a certain level of discipline from the user to use tags consistently, but they work well IME.
These days with ML tools it's easier than ever to assign relevant tags to URLs automatically, though I would personally not rely on them because of their propensity to hallucinate and not follow conventions.
And, like you say, web sites change and disappear constantly, so snapshots are important as well.
I've been meaning to try SingleFile/ArchiveBox/Wallabag for months now, but I'm currently too comfortable with Pinboard. Pinboard has been working great for me for many years, contrary to the negative buzz around it, but I don't have confidence that it will stick around forever. I just hope that I can export and migrate all my snapshot data as well...
For long-form content that I want to come back and read later, I find Wallabag is quite good. For making snapshots of large websites I use Kiwix.
I'm trying out using the Obsidian Web Clipper extension, which essentially does this (and using it for anything I'd previously have bookmarked).
I've been using Zotero for that, but I again wish we had a browser maker that was concerned about users, and felt their only goal was to give the user as much control and ability as they can handle, and as much protection as they ask for.
Sadly, Zotero and a lot of other software in that line is built with no thought to speed, I think more than a few thousand entries will make it impossible to open, browse, or search. This seems to be a commonality of all of this type of software; that it isn't really meant to be used.
Logseq, etc, are tools that help facilitate this, albeit in not plain text format, but close.
Being able to save a page also seems to be against the interests of publishers now - they would rather you revisit the page than have a copy. Substack for example seems to try to block the web clipper because you could clip paid articles and read them after your subscription lapses.
I hate it when (news) articles change without acknowledging the changes.
Edit: come to think of it, MediaWiki + a crawler bot would be near ideal for this.
[1] https://support.mozilla.org/en-US/kb/bookmarks-firefox#w_how...
The thing is - I just saved bookmarks, I never really utilised them ever, to find something, to go back to. I can remember once or twice and either I couldn't find anything among my bookmarks or the sites were long gone. I really don't think I personally had to consult my thousands of bookmarks (which I have now dutifully migrated to Raindrop of course, because why the hell not) in any useful sense ever. I paid for a couple of archiving services as well before realising "nah, I don't really need that, nor this recurring outgoing payment in my life".
So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
That reminds me of note-taking. There was a time when I used to do "note-taking exploration and research" and never really took any notes or, hell, even needed them. When I started note-taking, while I still keep an eye out for a decent app, I just pick a decent or half-decent note-taking app and I just take notes. Oh, backup and sync tools and services. Those too - there was "explore and research" and now there's "just use something damnit". "TODOing" to, yes! I am sure this tool (or philosophy? style? bookmarking architecture?) is very nice and novel.
This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
You're not describing a bookmarks issue. You're describing a personal organization issue, which is reflected on how you manage bookmarks.
You're voicing the exact same sort of complains often directed at todo lists. In fact, from your description you're implicitly treating bookmarks as ad-hoc Todo lists, and you're complaining your To-do backlog is growing.
Like others, you can blame bookmarks and Todo lists for your growing backlog of things you want to do but never get around to doing. Those are not the problem though, and only reflect a symptom caused by the actual problem.
> So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
You're describing a symptom of your problem. The fact that it extends beyond bookmarks is a telltale sign.
> This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
I believe you're expressing the same issues expressed by those who have trouble managing their task queue. Your problem reflects on bookmarks, on personal notes, on productivity software, etc. This means your problem is not bookmarks, or personal notes, or productivity software. It's something else that is reflected across tools and systems.
I have a huge bookmark collection, but I don't care if I saved a bookmark that I never opened. I configured my browsers to exclusively use bookmarks in recommendations, thus serving as an ad-hoc search engine of noteworthy links I visited or want to visit. If I don't visit any of the links, that's too bad. Why would it present a problem?
I never bothered to look at the schema of the bookmarks.html, because feels like it's worked the same for 20+ years. I used to care a lot about housekeeping the structure, but it doesn't really matter, as long as they're in the bucket the browser will use them for autocomplete suggestions, so...
Reading the author's description made me realize how unbookmark-like bookmarks actually are. The current implementations are somewhat akin to creating a list of books that you like at the library. It's not so much a pointer to the information you found useful, as it is a list of books you found useful. You still have to do some digging when you go back for the book. If the book is lost, you end up having a reference to something that you cannot obtain. And if you just add books to the end of your list, you still end up having to search through the list. The only way around that is to spend time organizing your list. It's no wonder why bookmarks are useless to so many people.
The author doesn't really solve the problem with bookmarks, except for one. The last one. By sticking a bookmarks file in a project directory, at least you're only searching through a list of bookmarks relevant to the project. If you are no longer interested in the project and delete it, you're also getting rid of bookmarks that you (hopefully) no longer need. It also addresses the portability of bookmarks. As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services. Look at moving bookmarks from one Firefox installation to another: you either use online sync, export to HTML to import from HTML, or import the database (which replaces your current bookmarks with the ones being imported).
There is widespread browser support for linking directly to text fragments[1] which makes it possible to link to arbitrary parts of documents even when the author hasn't marked up some nearby element with an id to target, like so: https://news.ycombinator.com/item?id=45047572#:~:text=You%20...
Unfortunately, Firefox provides no convenient way of creating such links, but Chromium has a "Copy link to highlight" context menu entry when you have text highlighted. Neither Firefox nor Chromium provide a convenient way to create a bookmark to a text fragment.
> As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services.
Both Firefox and Chromium support importing/exporting bookmarks from/to "HTML" (really, Netscape bookmark file format[2]).
Both browsers also provide the means to organize bookmarks into folders, and Firefox lets you add arbitrary tags to them as well. Alas as you say, the only way is to spend time manually organizing them. Automatically suggesting folders/tags (preferring ones you already have created) seems like an ideal use case for LLMs or other NLP tools. Ideally browsers would offer an option to save a snapshot of the page together with the bookmark, that would guard against link rot and enable full text search. We have the technology, it's really only a matter of improving the UI and linking the two features together. Too bad hamstringing adblockers, gimping sites that rely on XSLT, and implementing WebBluetooth or whatever has higher priority.
[1] https://developer.mozilla.org/en-US/docs/Web/URI/Reference/F... [2] https://learn.microsoft.com/en-us/previous-versions/windows/...
And — possibly to also literally keep them inside the browser’s default bookmarks/favourites whatever browser one uses. Not on some fancy service with AI and what not.
I try to read little I am not looking to apply, or be conscious it's for pleasure/interest
If I bookmark something, I consider it unread. If I read something, I make sure I bookmark and annotate it and tag it to make my mind more actively work with what I'm reading (and make it easer to find.)
The result? 10-15 years of every link I've ever saved, organized and annotated by me. Chronological, sorted, I can see what I was paying attention to chronologically, or by topic, and at any time search any of my highlights and notes.
This is the nice part because it isn't an AI tool, but maybe something that can feed into an AI tool quite nice. My curation, where relevant, as input.
Best of all, it just works. It's not heavy or tedious, anything that has my attention, gets my attention.
The one thing a text only approach will not solve is that URIs while universally defined will not perpetually stay online.
Diigo, and other tools like it allow you to save your own cache, or perhaps submit to a public cache that page so once it invariably goes offline, it doesn't.
There's lots of tools out there to help with this each person's way, I liked diigo.com, but lately think tools like logseq with a few basic plugins are offering a lot of promise to directly save a bookmark, whatever snippets are relevant, and they are always and instantly searchable.
Which means that spending 10 seconds tagging and categorising a link before saving it becomes really useful.
If I suspect even a little that the site or content might disappear (completely or behind a paywall) later, I use Obsidian's Web Clipper to save the whole text locally.
Everyone's workflow is different, but I personally do this often. I bookmark something interesting or valuable precisely because I might want to come back to it at some point. Sometimes I do, sometimes I don't, and that's fine. But when I'm looking for good software that does X, a good place to stay in an interesting location, or a good product in a specific category, my bookmarks are the first place I check. It sure beats relying on web search results and having to navigate around ads, spam, scams, astroturfed forum discussions, morally bankrupt SEO-hacked listicle sites, social media garbage, and the modern web dumpster fire.
Now I mostly keep two kinds of bookmarks: quick-access ones for work (like repos I contribute to or PR sections I need to check often), and then more organized notes for ideas, projects, or interests I want to revisit later. To make that easier, I use a little tool I put together (beavergrow.com) where I can group bookmarks into blocks and keep notes alongside them—it’s been handy for giving some structure without overcomplicating things.
The file it generates has:
<!DOCTYPE NETSCAPE-Bookmark-file-1>A bit similar to static web site generators today, only with endless security vulnerabilities.
javascript:window.prompt("Title + URL:",document.title+"\n"+document.URL);
Bookmarks.txt is a concept of keeping URLs in plain text files | Hacker News
https://news.ycombinator.com/item?id=45047572
to hn
The script will open the site https://news.ycombinator.com with Chrome. I also have a help screen that lists the URL's and their mnemonics. Again, note that this is a very short list.
I like working at a command-line so it's often faster for me to run something like this instead of typing the first few characters of a site name into a browser and waiting for the URL autocomplete to finish the URL for me.
Browsers have been doing an excellent job of managing bookmarks, you can tag and search for them from the address bar itself which is very convenient.
If you organise them you can even reference them from the codebase, or the documentation to avoid clutter. The format is simple and dumb enough so that a simple bookmark.txt can be converted into a dictionary, array that can be used in the program if some URLs are supposed to be used there.
It's not revolutionary by any means, but I have to confess that it didn't occur to me that's a great per repo documentation reference tool or per folder.
The KRunner plugin I use doesn't have a comprehensive Markdown parser, but it works great with the format I've been using. https://github.com/andrewrabert/krunner-markdown-bookmarks Ex:
# Bookmarks
## Code
- [GitHub](https://github.com)
- [GitLab - Arch Linux](https://gitlab.archlinux.org)
## Social
- [Hacker News](https://news.ycombinator.com)
- [Reddit](https://www.reddit.com)
I've solved the "non-curated" with Karakeep; especially its AI autotagging.
Now I have to figure out the next thing I want, which is a self-hosted service that
- saves bookmarks, duh
- saves "readability" style copies (no need for other stuff, I just need text)
- some decent method of collections beyond just tags
and the last thing that is very oddly rare
- publically accessible WITHOUT login (this is partly for publishing readings to classes I teach)
I can't find a sweetspot of all of these, may have to just build it myself?
This works for me because I don't try to sync the file. I don't browse on my phone much because I hate typing on one, and I keep work and home separated.
It may be time to find another metaphor for this second group: "postcards" might work better than "bookmarks".
Also, I sometimes add a snippet of text from the content to the title of the bookmark, making it easier to search afterwards.
I think that simple URL bookmarking is just wrong. It simply will not work for big bookmarking data sets. The key is using tags, and rating system, and automatic update which checks if URLs are even valid any more.
I also thought that we miss a killer RSS app.
That is why I created my own self-hosted app.
- it can store bookmarks
- it gather news through RSS
- it provides tags (I can search bookmarks by tags)
- it provides user ratings (I can filter using it too)
- I can filter, or order by link, date of publish, date of creation, etc. etc.
- It checks if links are rotten (and marks them)
- I can mark link to read it for later
- I can see how many times I have visited a link
- I can check 'related links' to jump to things I have jump before from this link
On the other hand, I am quite certain that I use it, because it is 'tailored for me'. I am not that interested in the looks. I know how it works
- https://rumca-js.github.io/search - demo search
- https://rumca-js.github.io/music - demo music
- https://rumca-js.github.io/bookmarks - demo bookmarks
- https://github.com/rumca-js/RSS-Link-Database - database of bookmarks
- https://github.com/rumca-js/Internet-Places-Database - link meta information
- https://github.com/rumca-js/Django-link-archive - main crawling engine using for all databases
1. Use Raindrop and its browser extension (or mobile app) for saving bookmarks.
2. Subscribe to the RSS feed of my new bookmarks.
3. Use Feedmail to send my bookmarks to me via email, with the full scraped article text included.
Email is a durable format with lots of powerful tools and easy automation. My bookmarks and workflow can survive the death of any of these services. I can tag, search, and read my bookmarks anywhere along with all of my other feeds, newsletters, and notifications. I can also easily forward an article I've bookmarked this way, or reply to self if I want to save my own commentary on it.
The specific problem I’m trying to solve is searching for seeds. I have about 20 different seed vendors bookmarked I want to search across all of their seed catalogues at once (without having to visit each site to search one at a time).
More broadly though I would love the ability to do custom / curated search. A search engine designed for searching collections of sites instead of the entire internet would be ideal.
It has been very useful. URLs are super easy to modify, super easy to share, super easy to open, add notes etc. Having a gist adds the ability to share a set or URLs and people can comment on them.
I could (if i want to) use github API and a browser extension to put bookmarks directly in a gist as plaintext
After some years of using this extension every day I decided to make it available for others as well. You can find it here, free forever: markbook.io This was just a side project I did in about 6 months of my spare time, definitely not a polished product that's trying to become the next billion dollar enterprise.
Another one (that does sync) is Floccus. https://floccus.org/
I'm part-way through getting the 90,000+ PDF files collected in this manner, analyzed by an LLM so I can .. query it about my own interests, I guess? ;)
I don't find that saving URL's is very productive - they are the dangling pointer of the web. Far better to have your own cache of docs to refer to imho ..
- tab separator instead of space. you don't need cut -d with tabs. tab is the default separator for cut.
- use xsel -b to capture clipboard data while adding. Copy the URL into from browser to clipboard. And then run on console
$ url.add hn bookmark script # all parameters are keyword tags.
The more mental effort put towards something, the easier it is to remember.
reading through this thread I’m seeing people use bookmarks to save dozens of new URLs per day, which is very surprising to me
I would suggest the date field be YYYYMMDD for sortability.
20250828 https://news.ycombinator.com/item?id=45047572 Bookmarks.txt is a concept of keeping URLs in plain text files (github.com/soulim) #bookmarking #textfiles`For those curious: https://studium.dev - built with Quartz (https://quartz.jzhao.xyz/)
(For those curious about why one needs so many bookmarks, similar to maybe some other people, I use the bookmark feature as a "like" to save articles/URLs I find interesting).
I switched to saving pages using SinglePage instead, that saves the current page as a single stand-alone HTML file. It loses the bookmark-like features, but I can sort those saved files easier in my file system to keep them like any other downloaded documents on various topics. Each file also by default has a comment near the top with its original URL, so it would be easy to write a script to find all of those and build something like bookmarks.html.
For my CTO newsletter I use raindrop.io to store interesting articles I encounter, export them to CSV (like bookmarks.txt), sort, filter and remove 90%, convert them to my own format, write my content and convert them to Markdown and then to HTML with Hugo.
[1]: https://learn.microsoft.com/en-us/previous-versions/windows/...
If you are motivated enough to write Apple Shortcuts a useful trick for ~Find type actions that overload is to Filter them into reassemblable pieces eg
action find items from reading list filter Title begins with A (then do B etc)
that this trick often works would be due to the internal nature of the Shortcuts implementation problems, so YMMV
the closest i've found is hypothes.is but i'd love to find an extension that maintains this information locally so that i dont have to periodically auth to a web service.
For frequent websites I can remember their TLD and navigate to the panel I wanted
For content saving a URL is not enough. Gotta save the full page or it might be gone very soon.
Each new day gets a header.
Here's Sunday, with blank lines added to get reasonable formatting on HN:
links from 02025-08-24:
https://archive.org/details/TheDesignOfSwitchingCircuits/pag... My childhood book on digital logic (including a little #electronics) as an #ebook. #hardware
https://en.wikipedia.org/wiki/High-intensity_discharge_lamp sodium-arc and mercury-arc lamps are also HID lamps. #hardware
https://archive.org/details/ge-glow-lamp-manual-1966/page/n1... scan of GE’s glow-lamp manual for neon lamps from 01966. #ebook #hardware #electronics #history
https://youtu.be/Nn5v59l2Xec?t=64 VEMAG brand double-screw extruder screw pump #mechanisms pumping M&Ms and ground meat. #video footage of the parts being washed.
https://www.radiomuseum.org/r/aleksandro_arz_51_arz51.html very pretty old radio
https://www.nature.com/articles/nmat2141 "Superlenses" to overcome the diffraction limit, #optics #paper from 02008
https://en.wikipedia.org/wiki/Electrical_length#Electrical_l... electrically long and short #antennas and loading coils and whatnot. #electronics #communications #radio
https://www.ornl.gov/publication/evaluation-power-fluidic-pu... #fluidics for pumping in #molten-salt #nuclear-reactors (just an abstract)
https://www.hopefulmons.com/p/in-defense-of-tech-trees trying to use "tech trees" to understand the #history of technological development
https://en.wikipedia.org/wiki/Fresnel_diffraction near-field #diffraction is "Fresnel diffraction" #optics
https://psi329.cankaya.edu.tr/uploads/files/Lewis-PernFascis... Was #Perón a fascist? #fascism #history #PDF #paper #toread
https://lwn.net/Articles/1030818/ #Treacherous-Computing for “confidential VMs” #privacy despite #Linux #virtualization #toread
https://news.ycombinator.com/item?id=44950482 discussion of alternative ways to run graphical apps inside #Docker, including maybe drawing in a web browser
https://news.ycombinator.com/item?id=44991638 my post about how #Rust’s approach to handling #Unicode introduces unnecessary bugs into command-line programs
https://www.youtube.com/watch?v=WApL1EL2GMk #Frondizi speaks about #Perón. #history #Argentina #video
https://github.com/3b1b/manim Manim math #animation software for morphing equations and plots and stuff into each other, using FFmpeg, OpenGL, LaTeX, and Pango
https://nvmexpress.org/wp-content/uploads/September-2020_NVM... in ZNS #zoned-storage a zone must be a power-of-2 number of LBAs, unlike in ZAC and ZBC #PDF #toread
https://docs.kernel.org/filesystems/f2fs.html "f2fs" is the #Flash friendly #filesystem for #zoned-storage. “F2FS is a file system exploiting NAND flash memory-based storage devices, which is based on Log-structured File System (#LFS). The design has been focused on addressing the fundamental issues in LFS, which are snowball effect of wandering tree and high cleaning overhead.”
———⁂———
Here's today, ten years ago, so you can see how my bookmarking style has developed:
links from 2015-08-28:
http://www.excamera.com/sphinx/article-j1a-swapforth.html A free-software #Forth operating system running on the J1a open-source CPU running on a Lattice #FPGA thanks to Project #IceStorm’s reverse engineering and the resulting free-software #synthesis and programming toolchain, running on the Lattice iCEstick evaluation board with 8K of RAM. #J1 #hardware
https://www.youtube.com/watch?v=uRVaLUQUmA8&list=PLACB124F79... A 2009 #video lecture series on #mecheng manufacturing processes by “nptelhrd”, Prof. Inderdeep Singh at IIT Roorkee. Some problems with the audio, seems to be a bit clipped. However, the video is entirely a talking head and a bunch of PowerPoint slides, so the only way you would watch this video to learn about powder metallurgy instead of reading a book is if you are dyslexic or have a beard fetish. At least it doesn’t have shitty elevator music.
http://minerals.usgs.gov/minerals/pubs/commodity/ The USGS gives #pricing information on a variety of mineral #materials, including some historical stuff. #minerals
http://minerals.usgs.gov/minerals/pubs/commodity/titanium/mc... #Titanium #pricing in the US has gone from US$9.62/kg in 2010 to US$11.20/kg in 2014. Nearby pages use dollars per pound, but the prices per kg show a sharp reduction from its peak of US$17.28 in 2005, but it had a low of US$6.50 in 2003, US$9.70 in 1995, and US$8.26 in 1992–3. This means that the FFC Cambridge Process is not in production yet.
http://pubs.usgs.gov/sir/2012/5188/ A #PDF with a bunch of metal #materials #pricing for 1970–2010. Comprehensive, covers nearly all metals and some semimetals. Unfortunately at least some prices use folk units like pounds instead of SI.
http://rebeccasolnit.net/essay/a-rape-a-minute-a-thousand-co... Rebecca Solnit talks a bit about #rape and #intimate-partner-violence. #feminism
https://www.federalregister.gov/articles/2015/08/06/2015-184... The FCC is proposing to outlaw #free-software for wireless firmware.
https://codegolf.stackexchange.com/questions/4732/emulate-an... Programmers competing to do Intel 8086 (subset) #emulation in different languages, including a C program in 348 lines.
———⁂———
Perhaps surprisingly, of those 8 links from ten years ago, only one has really linkrotted; unsurprisingly it's Solnit's essay. But it survives in the Archive. The titanum PDF redirects to the general titanium Mineral Commodities Summary page.
Emacs full-text search is surprisingly often effective at finding relevant information. Grep (or M-x occur, its moral equivalent) works even more often. F6 opens the URL in my browser. Emacs C-s can search those 12000 bookmarks faster than I can type.
At some point in the past I wrote a Chrome extension that would show me all the links for a given hashtag with their descriptions (formatted as Markdown), with the hashtags being links to other similar hashtag pages, and also let me bookmark pages from within the browser interface, but I don't know where it is now. It wouldn't work with Manifest v3 anyway.