What am I getting myself into? Maybe it was a combo of @m_ashcroft's enthusiasm, and some interesting discussions on an Obsidian forum today... I still have a pretty high confidence level that Obsidian will never be my daily driver (but not 100%)...
However, I'm really frustrated by Roam missing some essential features, and Logseq still having too many performance problems... So let's try something totally different :)
Also, there's an insane amount of cool plugins and good thinking in the Obsidian community,
which becoming more familiar with Obsidian might help me learn more from. And I like the premise of the course, and the style - also going to be taking notes on pedagogy, onboarding etc.
@rroudt might want to get inspired for a self-guided course on Logseq.
First thought: The live preview is really good - seems even better than Roam/Logseq for many things (hiding Markdown in the block your editing but still easily editable)
Thought on the pedagogy - I love that there's lots of text to be read, interspersed with short videos. @cortexfutura I'd love to see much more writing in the future, courses that only consist of video are really exhausting. Also excellent interactive format - now edit this, etc.
Interesting to see different ways of formatting notes - putting note title on top as header, using headers to structure content - seems way less flexible than an outliner (oops, I made my sections h2, but now I need a category above). Seems clumsy to me...
Of course I know that you can use Obsidian more like Roam (outliner plugin etc), but also useful to learn the "native"/"idiomatic" way (even though I know there are many approaches).
I will say - this course looks visually awesome. So pleasant to navigate/read. Never feel lost. A sense of calmness.
Many nice themes foe Roam, but even with document mode etc, don't think you could get to anything similar... For sharing/presentation etc, definitively ++.
Tiny thing, but it's interesting that the open new page/create uses enter to select the top choice, but shift+enter to create a new page. This means both options are a keypress away, whereas with Roam, you have to arrow down and enter to select the existing page. Speed matters.
The small keyboard legend at the bottom is also unobtrusive, but great for discovery - does everyone know that shift+enter in the Roam dialogue opens the page in the sidebar?
(What's better - dropdown or modal? Possibly Roam dropdown let's you keep more of the page visible).
The toggle pin, which can be toggled with Shift+Cmd+E keeps the page you are on, and all links automatically open to the right. Tiny detail, but really neat.
One thing I've wanted: to be able to select multiple pages from the autocomplete. Say I search Obsidian, and see 3 relevant hits - I'd like to be able to shift-click on all 3 of them to open.
(Of course I can do a search, rather than using Cmd+O, but if I'm already there...)
Obsidian quick-switcher does partial matches, so nimil will match Nick Milo (I have a lot of Nick's!), whereas Roam throws up its hands. No nimil's here, go make one. :)
Although this doesn't work everywhere. If you have a tag called #on/pkm, typing #onpkm will not autocomplete.
Talking about tags, one big difference with Roam is that Obsidian treats tags and pages totally differently. Writing [[peter]] creates a page called peter.md when clicked on (before created, link has different color). Whereas #peter only opens the tag panel...
In Roam I used these interchangeable - mostly to distinguish inside/outside text, like
- Met [[Peter]] to discuss his use of [[Logseq]] #meetings
- Need to plan my [[meetings]]
In this case, the two "meetings"s both refer to the same page, and generate the same kind of backlink.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Vision: all of my thoughts and readings are well indexed, and show up in backlinks where you'd expect them.
Actual case (10 min ago): Hm, this reminds me of the memory researchers who compared showing someone many paintings by the same painter, vs paintings by different painters
That was from a really interesting podcast, I remember I took copious notes. I also remember which street I was on, and what the weather was like, when I listened!
But why isn't it showing up in my [[memory]] backlinks? I don't have that many...
Which podcast was it? I heard about that podcast on another podcast, called Oli something... ah, @ollie_lovell , and the podcast was @mrbartonmaths "Mr Barton Math Podcast".
Using @RoamResearch after exploring @logseq, I desperately miss page tags. I want to tag the entire contents of this page #transfer, so that I can later search for eg. (and [[transfer]] [[perception]]). I refuse to put a bullet with #transfer on top and nest everything below.
The infuriating thing is that in theory, top-level attributes in Roam should be able to play this role. However, you can't query for them. I was tweeting about this two years ago - no improvement.
The founder of @logseq, @tiensonqin recently gave a long interview to the Chinese podcast ByteTalk (@lzzy@laike9m). I thought I'd share some key points here for all of the non-Chinese speakers. Hope I caught it properly, 请多多指教!:)
Tienson used to work remotely for a Canadian company, and used Emacs and org-mode. He began developing Logseq because he wanted something that he could use on iPad/Mobile, and dreamt of a tool that his daughter could one day use to learn and grow. Initially built it for himself.
He began building a tool based on org-mode and Workflowy, only came across Roam Research 5 months into the project, but really liked it, and took many inspirational ideas from it.
The community really grew out of Discord, after half a year they had almost 2000 users there.
For people curious about the Roam API and confused by the syntax, or interested in why Conor went with Datomic/Datascript and not a traditional database, this older talk by Roam developer @mark_bastian is a great overview.
He gives great examples using Spiderman of how even modeling something fairly trivial in SQL is much more complex than in Datomic. But the real kicker is when you're trying to interrogate the data to find recursive relationships.
Right now the Roam data model (at least that's exposed to developers) is just about pages, blocks, and children with tags. Already you can see how finding the page containing a block with a certain tag etc is useful.putyourleftfoot.in/introduction-t…
Somehow I’m watching a Netflix movie about an Indonesian woman in Azerbaijan falling in love with a local boy who is obsessed with Indonesian shadow theatre. I love it.
The bizarre thing is that I lived in Indonesia 12 years ago, I’ve actually read the novel the protagonist of this film teaches in her class, and I once was close enough to Azerbaijan to see the border, but never crossed.
Before this I went for a walk and listened to Indonesians on clubhouse discuss coffee manufacturing and Italian Robusta preferences. It’s weird how these global monoculture apps are at the same time bringing more diversity -
Fantastic work! So cool to have been able to contribute in a small way to this, but full disclosure, I would have really struggle with the actual algorithm that calculates the various table layouts - very powerful!
@conaw Would be really nice if you could export a Clojure component that lets us render links/tags in a proper way (allowing us to shift+click to open in sidebar etc), so that links can be rendered properly in these tables.
Also nice if we could get a nicer way of calling these, rather than block-reffing... Can a render plugin register to an XRef like {{xtab}}? Or perhaps we could abuse the template system, to get it to show up in the pulldown menu?