4 ms·

The graph seems more like a gimmick, but the automatic backlinks are nice. Org mode is a really good text-file based solution for notes, especially with tools

The graph seems more like a gimmick, but the automatic backlinks are nice. Org mode is a really good text-file based solution for notes, especially with tools like org-rifle: https://github.com/alphapapa/org-rifle https://github.com/alphapapa/org-rifle The other day people wrote in an other thread they use grep for their text notes. But what if you have notes with longer, multiline texts? With org-rifle the input words can be anywhere in the multiline note and they can be in any order.

karlicoss 7y ago
I've got a massive org-mode database, and in 90% of cases I'm using ripgrep (helm-rg) to search over it. In addition it searches in chat logs and other non-org mode files. Remaining 10% of searches is probably org-ql (e.g. tag search). Tried org-rifle too, but also wanted searching within the contents, so ended up using ripgrep. I describe my setup here: https://beepb00p.xyz/pkm-search.html#other https://beepb00p.xyz/pkm-search.html#other
secstate 7y ago
I'll say that I've been using org-mode and it's agenda to manage todos and take notes for a long time, and Roam is whole other animal. The problem I've had with Rifle is that words can mean different things in different contexts. Rifle finds every time I used the word "depression" regardless of how I used it. Roam gives you any easy way to add just a little extra symantics to your notes that dramatically increases the signal in the backlinks, and astronomically improves the graph view.
- dmortin 7y ago
  Could you show an example of those "little extra symantics" which helps you with this?
  - secstate 7y ago
    For sure! So if I'm reading an article about the fall of the Roman Empire, and I learn about how modern scholarship does not talk in terms like "fall" much anymore, but rather civilization transformation, I might make a new stub article titled `civilization` and then another one for `civilization transformation` Now, if I read a story about Asoka's Empire in what is now India, and I read about how it underwent a collapse, I might tag it with `civilization transformation` as well. Another story about the nature of modern civilization and the travails of people who treat their kids as a personal project to "build" as opposed to a living being to nurture and encourage, I would tag that with `civilization` but not `civilization transformation` Ripgrep'ing my org files for "civilization" would turn up all three of these backlinks. Grepping does not understand context. But I understand context. While `civilization` is part of `civilization transformation`, I don't want `civilization transformation` surfaced for every search of `civilization` This is a rather contrived example with some obvious issues, but I suppose a good analogy would be that you can drag fish for scallops and get every other living thing off the bottom in your net, increasing the work to clean it out. Or you can dive for individual scallops and take longer but have less work to do when you surface. Both methods have merit, but one provides a much cleaner signal. Practically speaking, I USE my Roam note's backlinks. While I could grep my previous notes, in practice I never did because it was too much work.
    - dmortin 7y ago
      Thanks. I was curious why org-rifle is not sufficient for search, but you just describe searching by tags which org-rifle can also do, or searching within a tag.
    - b3n 7y ago
      If you are using backlinks, why not just grep for `[civilization]]` or `[[file:civilization.org]`? I actually have a binding in Emacs to run ripgrep on all files in the current directory, which have a link to the current file. It's very simple, and runs in a split second even with 1000s of files.
jarvist 7y ago
grep -B / -A / -C [lines Before / lines After / lines of Context] is very useful for when you have multi-line text you want to extract.
- dmortin 7y ago
  The point is searching multiple terms any order in a note. With org-rifle I don't have to think about the number of context lines or anything and I just type the terms and it finds the notes which have all of those.

karlicoss 7y ago

I've got a massive org-mode database, and in 90% of cases I'm using ripgrep (helm-rg) to search over it. In addition it searches in chat logs and other non-org mode files. Remaining 10% of searches is probably org-ql (e.g. tag search). Tried org-rifle too, but also wanted searching within the contents, so ended up using ripgrep. I describe my setup here: https://beepb00p.xyz/pkm-search.html#other https://beepb00p.xyz/pkm-search.html#other

secstate 7y ago

I'll say that I've been using org-mode and it's agenda to manage todos and take notes for a long time, and Roam is whole other animal. The problem I've had with Rifle is that words can mean different things in different contexts. Rifle finds every time I used the word "depression" regardless of how I used it. Roam gives you any easy way to add just a little extra symantics to your notes that dramatically increases the signal in the backlinks, and astronomically improves the graph view.

dmortin 7y ago
Could you show an example of those "little extra symantics" which helps you with this?
- secstate 7y ago
  For sure! So if I'm reading an article about the fall of the Roman Empire, and I learn about how modern scholarship does not talk in terms like "fall" much anymore, but rather civilization transformation, I might make a new stub article titled `civilization` and then another one for `civilization transformation` Now, if I read a story about Asoka's Empire in what is now India, and I read about how it underwent a collapse, I might tag it with `civilization transformation` as well. Another story about the nature of modern civilization and the travails of people who treat their kids as a personal project to "build" as opposed to a living being to nurture and encourage, I would tag that with `civilization` but not `civilization transformation` Ripgrep'ing my org files for "civilization" would turn up all three of these backlinks. Grepping does not understand context. But I understand context. While `civilization` is part of `civilization transformation`, I don't want `civilization transformation` surfaced for every search of `civilization` This is a rather contrived example with some obvious issues, but I suppose a good analogy would be that you can drag fish for scallops and get every other living thing off the bottom in your net, increasing the work to clean it out. Or you can dive for individual scallops and take longer but have less work to do when you surface. Both methods have merit, but one provides a much cleaner signal. Practically speaking, I USE my Roam note's backlinks. While I could grep my previous notes, in practice I never did because it was too much work.
  - dmortin 7y ago
    Thanks. I was curious why org-rifle is not sufficient for search, but you just describe searching by tags which org-rifle can also do, or searching within a tag.
  - b3n 7y ago
    If you are using backlinks, why not just grep for `[civilization]]` or `[[file:civilization.org]`? I actually have a binding in Emacs to run ripgrep on all files in the current directory, which have a link to the current file. It's very simple, and runs in a split second even with 1000s of files.

jarvist 7y ago

grep -B / -A / -C [lines Before / lines After / lines of Context] is very useful for when you have multi-line text you want to extract.

dmortin 7y ago
The point is searching multiple terms any order in a note. With org-rifle I don't have to think about the number of context lines or anything and I just type the terms and it finds the notes which have all of those.