The future of Logseq - Sneak peek at the DB Branch
Вставка
- Опубліковано 9 лип 2024
- Logseq is hard at work to improve the core functionality, but why is it taking so long and what can we expect? I take a dive into the code to get a sneak peak on a possible future.
Table of Contents
00:00 - Start
09:51 - DB Export Thoughts
11:58 - Export options
13:47 - Demo of the DB version
25:38 - Export Options
26:29 - Final Thoughts - Навчання та стиль
Thank you for this look into the future of Logseq. I’m super excited for it.
If there was a date for it I would be counting down. But until then I'll just think up of things I'm going to test when it hits. Excited :D
Starting to look a lot closer to what capacities and tana does - really excited for this release.
I'm glad it's getting these features as it means I can stick to owning my data but the system will keep me a bit more in check. I'm pretty chaotic and templates only bring me so far. Still, it's going to be a few releases before this one gets a full release.
Insane. I'm so hyped for this to hit the released versions
This is soooooo helpful!!!!! Thank you!!!
Awesome video! Thank you.
Always glad to hear!
Super Nice🤓
Thanks a lot for the sneak peek.
Glad you liked it, was really curious as well and learned so much making this.
Great view into the future of Logseq. Thanks a lot, Bas! I use Logseq and want to use it as an personal information-storage. The new developments make that goal more reachable. I will stay tuned!
Really looking forward to doing more and more information storage in Logseq, structure and control over my data? I'll take the entire stock!
Great work! I am a heavy user of Logseq, utilizing it daily for note-taking, personal knowledge management (PKM), and task management. As my notes expand and accumulate, organizing and retrieving information when needed becomes increasingly challenging. A database (DB) version of Logseq would be ideal for me. Thank you for sharing your work and demonstrating what the Logseq DB version could look like. I hope you can contribute to the Logseq team and expedite the development of the DB version. I can't wait to see the DB version released.
Please note it's not my work, this is all done by the Logseq team and people that know way more clojure then I ever will I'm afraid. But I hope that I take a bit of load of the team helping promote all the good work and doing a back and forth on the social media questions there.
Thanks for showing this. I appreciate Logseq for being free, local storage and open source, but I don't get the sense that they have a vision for their product beyond catching up/copying the way other apps did things (Roam and now Tana). But I hope I'm wrong, and I hope with this new DB version we start to see their own vision for this product/category and where they'll innovate. Their whiteboards are my favorite canvas/whiteboard feature of any note-taking PKM app so far.
One feature I didn't show and after it got explained to me on Twitter got me really exited. You can tell fields to add their status icon to the beginning of the line. That way you can build your own quick single line status overview. For example, with tasks I now have a checkbox, icon that's colored to show urgency and an icon to show effort. And that one feels new, I at least haven't seen it in other apps. And that gets me excited.
But in the end PKM has a finite amount of parts for infinite information so I rather have good implementations of the best ideas then features to be unique that I never use.
Not only a lack of vision, but a lack of care for paid Sync users and its many data loss issues. There are TONS of data loss tickets being filed, with no resolution for the corrupted data. I wouldn't trust that company with my money or data.
@@lemonline3719 Yeah, good point. It is strange, I’ve noticed more errors with saving pages lately than I did when sync first came out. While the aforementioned things had me worried about Logseq and looking at other apps, the saving/syncing errors made me stop using it altogether lately except for checking/migrating things I already added. I hope they manage to get things smoothed out
Thanks Bas for the overview of the new Logseq. The lack of news and updates indeed left me wondering about the direction the project is going. As much as I love the current Markdown storage, I am also excited to see how the change to a DB will be beneficial in terms of performance and what kind of new plugin will come from it. Cheers and I really appreciate all your videos.
That exact reason was why I was diving into it, while I could mostly live with just a Markdown export I'm glad to know they are going to look at 2 way sync to Markdown giving us the best of both worlds. Though I would rather have a text based solution that was more block based like yaml. Working on a video about that topic.
Thanks for the video, it would be nice to see a query of previous created tasks, sorted let say by project, or arranged ina kanban. Which I think is the final goal of adding proprierties
Sorted by and Kanban are high on my wish list. And I'm sure those will appear quickly enough in the base app once this rewrite is done because they are natural progressions on having the data in structure.
If they can use the structure so I can do an instant quick add on my phone, that would change my day to day workflow in a huge way and I'll probably be making a 10 min video with me just looking happy and ranting about it then.
Still not convinced this is something I would like. Fortunately as you said in the beginning, you can keep using existing versions. I will have to see what the product looks like when it is officially released.
I'll be for sure checking the backend and import/export options as having my data in an easy to read format is one of the core selling points for Logseq.
Yes, I didn't expect to move away from Markdown and complicate interaction with many dialogues. I thought they'd be only indexing into databases SQLite database, which would make a perfect sense.
Finally a great inside view into the DB version. I have to say if you are just a user without any IT background it still is a lot to swallow. I will have to make a whole new tutorial series to get a grip on that myself.
The only question now is if one should keep on feeding the markdown backend or wait for the DB version. I still have hundreds of book-notes to put inside my logseq. And it would be a shame if i have to do all the work again when adding the DB-Versiom features shown in your video.
I would keep going, the main thing to keep in mind is to make sure to use the tags:: property in pages as that will translate to the classes later on. This is why all my books that have a page now have tags:: book in them.
We will have to wait till the full release to see how easy it is to migrate. I got tons of notes so hope it's done well.
This gives me hope that the time I've spent in Logseq won't be wasted. The lack of proper typing is very limiting right now, and templates are a hack job at best. Good to see these things are being worked on. That DB made me cry a little though.
One of the reasons I made the video, because you could say I have quite the stake in Logseq with my channel. Good news though, Logseq team confirmed that they will work on 2 way sync to Markdown once the DB is stable. I hope it's manual with a keyboard shortcut though as I would hate for it to still fight with files on disk like it is now. And pushing a button would be a small price to pay for speed and stability
@@toolsontech oh yeah, I'm all for using a DB instead of files, I just meant those EDN files inside of a single table looked weird. And I've often encountered Logseq going out of sync with what's actually on the disk, especially when a remote sync happens, so using a proper DB would hopefully prevent that too. I think syncing DBs over the network is a solved problem.
The EDN initially got me bummed a bit but since then I realised it gets the main goals done. Not ideal but a good first step.
Glad I came across you. I was looking into Logseq deeper DUE to the fact that it seemed to have a little stability in dev (some call it lack) and wasn't pushing the plugin death matches with other note apps. Just when I get a good grasp on a full logseq to rdf mapping and llm inferencing with owl2 contextual RAG they go on and reformat... Its getting harder and harder to keep a grep on things nowadays. All I want to do is take notes and let my robot army manage the rest... is that so hard (its always FB profile complicated due to things like this)? Plain text triple blobs are starting to look like the wave of my future. Tool change velocity (with AI) is about to go nuclear.
Think sticking with the markdown version will get you that stability you seek. Does match my feeling that the core needs to be good and not change too much. Really hope that after the DB release the focus will be on speed and stability and not to much in new features.
Thanks for the preview! Did they mention any ETA for this? I'm looking forward to it... IF it comes with good export options or a way to turn DB-based graphs into Markdown based ones.
No ETA, it's the question burning on everyone's lips. Since the UX is getting polish I do hope that the closed beta starts soon as that would be one of the last steps before open beta. Before the biggest risk, going official where you want to be 100% that people won't lose data.
I'm keeping my fingers crossed at least!
Amazing content. In future videos, you could explore how the new "OOP-like" functions of Logseq compare with Anytype, as the "types" thing is one of their main "selling points". Also I would be interested in a comparison with TiddlyWiki. Unfortunately, the latter doesn't get as much recognition as it should in the PKM space imho.
It's been a hot minute since I last played with Anytype, my memories on it is that their types is an awesome idea but hard to understand initially. At least for my brain. TiddlyWiki is awesome and used it for a lot of DND campaigns, agree that people are missing out in the PKM space.
Anytype is great, but for me it's missing the key features of hierarchical bullet-based notes where you can embed/link to specific blocks instead of the whole page, blocks inheriting parent's tags, and proper back links to specific blocks.
I'm choosing Logseq
Would this change improve the sync features? This update looks cool and useful, but really hope this helps Logseq sync to be on par with other notetaking apps.
Yes, it will make a huge difference because now Logseq tries to merge changes on a best effort, in the DB version because it knows the time of every line it can merge files much easier.
Great stuff ahead. So its going the route of Tana, a bit more at least, right? No "Notion AI" kind of integration yet, (build in, no addons?)
Tana with offline and a mobile app.
I think the supertag idea from Tana is so good it instantly was clear that should be the new standard in any app that works like this. The AI part is from what I can tell currently on hold and I think that's the right move, instead of following the hype work on the foundation first. I do expect it in the future but it was confirmed that AI would always be a feature flag and that you could bring your own engine. Both I think essential for the current target audience of Logseq
So, there will be no "Pages" folder in future? Also, please comment on whether "Export as Public Pages" will be there?
I'm confident that Export as public pages will still be there, they also mentioned that work will be done for Markdown sync and you can pick between DB or Markdown backend if needed.
would it be possible to import the properties from the markdown logseq version to the new db version with the new property styles, or is it necessary to build new properties then?
From what I heard you will be able to import properties and there's switches to decide what to do. It's still in development though. There will also be a difference between links (brackets) and tags (hashtag), the last one is meant for classes so something to keep in mind already when writing notes
This is great..when is this version going to be released ?
No official date yet, some say it's almost ready for closed beta others say it's half a year out. My estimate is that we will get a beta in the next 2 months and then full release is very dependent on feedback from the testers.
Hi! Is there no query functionality in this database version? It looked quite similar to Anytype app.
Query was still the same when I was testing except for much better auto completion. From what I got on the discord there are plans to make it much more functional once the DB backend is there.
Will I still have to fight my way thru markdown codes (eg, ***bold-italic***) while editing text? I hate that. Now if I have to work with much formatted text I start with RemNote and copy-paste into logseq because RemNote hides the markdown while editing.
Don't think that changed a lot, while you can use shortcuts like ctrl+b and ctrl-i it will still show all the markdown. Personally I like that but I can fully imagine it's not for everyone.
thanks for this. this to me feels slower and worse (not from a tech perspective but workflow).
e.g. use case for a meeting during the working day. one of many, every day.
the DB workflow:
hashtag it
go into properties, tab into every one, fill in the gaps
you'll end up filling in that project property for example so many times
vs now
just use the outliner hierarchy/ indentation with project name
each meeting is a row with the persons participating, then indent for the content
that way, it's super fast, clean, and still fully retrievable, since Logseq shows all the parent blocks (so I know exactly where those meetings below), with no extra clicks.
unless I'm missing something?
Adding data to a structure will make things slower, but it's optional and should be build around your needs. Most of the work should still be done by just using the hierarchy, but a few things might be nice and I'll definitely will be experimenting to find out what is useful as a work method day to day and where it's just getting in the way of getting results.
Is it possible to use already via web browser on a phone like a pwa? Does it have a responsive design? Is there already a way to run this in a docker container?
Did a quick test on my mobile and it seems fully responsive. There's probably some bugs in there but first look is good.
No idea if it's already ready for a container. But I see no reason why the dev branch couldn't be added to a Dockerfile for self hosting. Keep in mind though that by design the data is still local on your browser so syncing would still apply.
@@toolsontech Thank you
i loved using logseq until my vault became a few hundred pages but i can't use a pkm that takes 5 sec each interaction on a high end pc. the whole thing feels like an overengineered backend with no frontend.
That sounds weird, it's plenty fast on my no high end laptop with 2000+ pages. Though the whole electron stack makes it ask way more memory then it should. Something the DB version could solved since it no longer needs to keep everything in memory but just load whatever is open.
Best ever app but they should encryption as well for their notes, like trilium has done.
Totally agree that notes should be encrypted at rest as well.
Honestly, this is not what I expected. I thought data will stay in Markdown, but all overhead indexing would go into databases. Not sure about this step.
If they get 2 way syncing working it might still be like that. As you could use markdown as a single source of truth. Just really hope that the sync will be better with error handling then what happens now.
@@toolsontech I sync with my Koofr.net cloud and it works perfectly so far. I have Google cloud too, but it has no meaningful and stable sync tool for my Linux PCs and laptop.
Things are a bit oversimplified to be black vs white here. There are techniques to augment the current approach with a separate index (could still keep the markdown as source of truth for part of the data), or use mmap, to alleviate some of the mentioned concerns. I find it weird the logseq team doesn't publicize design documents. To your point this would increase negative noise, but could also allow community members to actually participate in design (i mean technical design) and development.
It's mentioned that the goal is to get 2 way MD sync working. And that would give us the best of both worlds, stable day to day usage and a 2 way sync that could ask smart questions to avoid data loss. And if the only price to pay is that I need to hit a hotkey to sync up, that I could live with. (Though an automatic sync at startup/close would help)
On the Discord the frontend design has excellent conversations, but this isn't the case for the engine. My theory is that the core team is more developer than public relations and as such spending time sharing data designs and going through all the feedback would take too much focus away.
@@toolsontech that sounds like a very generous and optimistic take.
@@HMijailAntonQuiles I''m saving my snarky and cynical side for when I need to evaluate if they still adhere to their privacy goals once this release is out. :)
@@bgrolleman pls share your honest opinion
So is this a second branch or will the markdown version basically be burried after the DB release?
This second branch is the current Logseq version + db features, you can still use markdown graphs if you want on release.
I'll be keeping a close watch on the Markdown version as the sceptic in me thinks that will get less love over time. Let's hope I'm wrong.
@@toolsontech Thank you! I hope you are wrong, too. I kinda fell in love with markdown. Even with all it's potential performance issues i think for most people the clarity of markdown is just the thing that keeps it easy and understandable.
I hope this turns into a single app version that has both the Markdown and DB "modes", instead of two branches that have to be maintained (and can be abandoned) separately.
Will there also be a web version without the need to install the Logseq App? thnks
I can't confirm, but the demo I was doing was effectively a webpage so I don't see why not.
@@toolsontech let’s hope so👌🏻
Is that a panic theme?
Within Logseq? I mostly use the core themes and purple accent color, but over time I think a lot of old CSS snuck in there.
That's unfortunate as the fact that logseq keeps just markdown files was one of the reasons I decided to use it.
It's confirmed that they will work on a two way sync and that you can keep the old markdown setup. The latter will miss some of the more advanced functions but the core will remain the same
"Peak" following "sneak" may be the most frequent misspelling on the internet.
Yeah, easy to make since peak is also a valid word so spell check won't warn you. Fixed the title, thanks for pointing it out