diff options
Diffstat (limited to '')
| -rw-r--r-- | 2024/talks/p-search.md | 32 |
1 files changed, 16 insertions, 16 deletions
diff --git a/2024/talks/p-search.md b/2024/talks/p-search.md index 0827c44d..4d2b3fd1 100644 --- a/2024/talks/p-search.md +++ b/2024/talks/p-search.md @@ -60,10 +60,10 @@ Code: <https://github.com/zkry/p-search> - Q: Do you think a reduced version of this functionality could be integrated into isearch? Right now you can turn on various flags when using isearch with M-s \<key\>, like M-s SPC to match spaces - literally. Is it possible to add a flag to \"search the buffer - semantically\"? (Ditto with M-x occur, which is more similar to your + literally. Is it possible to add a flag to "search the buffer + semantically"? (Ditto with M-x occur, which is more similar to your buffer-oriented results interface) - - A: it\'s essencially a framwork so you would create a generator; + - A: it's essencially a framwork so you would create a generator; but it does not exist yet. - Q: Any idea how this would work with personal information like Zettlekastens? @@ -83,7 +83,7 @@ Code: <https://github.com/zkry/p-search> combine those into one author? - A: Currently using the git command. So if you know the emails the author have used, you can add different priors. -- Q: A cool more powerful grep \"Rak\" to use and maybe has some good +- Q: A cool more powerful grep "Rak" to use and maybe has some good ideas in increasing the value of searches, for example using Raku code while searching. is Rak written in Raku. Have you seen it? - [https://github.com/lizmat/App-Rak](https://github.com/lizmat/App-Rak) @@ -92,8 +92,8 @@ Code: <https://github.com/zkry/p-search> to include to have a better search. - Q: Have you thought about integrating results from using cosine similarity with a deep-learning based vector embedding? This will - let us search for \"fruit\" and get back results that have \"apple\" - or \"grapes\" in them \-- that kind of thing. It will probably also + let us search for "fruit" and get back results that have "apple" + or "grapes" in them \-- that kind of thing. It will probably also handle the case of terms that could be abbreviated/formatted differently like in your initial example. - A: Goes back to semantic search. Probably can be implemented, @@ -132,9 +132,9 @@ Code: <https://github.com/zkry/p-search> - A: You can search for whatever you want. No hardcoding is possible for anything (file, directories, tags, titlese\...). -- Q: \[comments from IRC\] \<NullNix\> git covers the \"multiple - names\" thing itself: see .mailmap 10:51:19 - - \<NullNix\> thiis is a git feature, p-search shouldn\'t need to +- Q: \[comments from IRC\] \<NullNix\> git covers the "multiple + names" thing itself: see .mailmap 10:51:19 + - \<NullNix\> thiis is a git feature, p-search shouldn't need to implement it 10:51:34 - \<NullNix\> To me this seems to have similarities to notmuch \-- honestly I want notmuch with the p-search UI :) (of course, @@ -147,13 +147,13 @@ Code: <https://github.com/zkry/p-search> could be adapted here to with the SQL database as source. - Q: You can search a buffer using ripgrep by feeding it in as stdin - to the ripgrep process, can\'t you? + to the ripgrep process, can't you? - A: Yes you can. But the aim is to search many different things in elisp. So there is a mechanism in psearch anyway to be able to represent anything including buffers. This is working pretty well. -- Q: Thanks for making this lovely thing, I\'m looking forward to +- Q: Thanks for making this lovely thing, I'm looking forward to trying it out. Seems modular and well thought out. Questions about integreation and about the interface - A: project.el is used to search only in the local files of the @@ -173,7 +173,7 @@ Code: <https://github.com/zkry/p-search> - A: Retrievel augmented generation: p-search could be used for the search, combining it with an AI to fine-tune the search with a Q-A workflow. Although currently no API. - - (gptel author here: I\'m looking forward to seeing if I can use + - (gptel author here: I'm looking forward to seeing if I can use gptel with p-search) - A: as the results are surprisingly good, why is that not used anywhere else? But there is a lot of setup to get it right. You @@ -201,7 +201,7 @@ Code: <https://github.com/zkry/p-search> - A: The code currently is dumb in a lot of places (like going of all files to calculate a score), but that is not that slow surprisingly. Elisp enumerating all files and multiplying - numbers in the emacs repo isn\'t really slow. But if you have to + numbers in the emacs repo isn't really slow. But if you have to search in files, this will be slow without relying on ripgrep on a faster tool. Take for example the search in info files / elisp info files, the search in elisp is almost instant. For @@ -212,13 +212,13 @@ Code: <https://github.com/zkry/p-search> - Q: When do you have to make something more complicated to scale better? - A: I do not know yet really. I try to automate tasks as much as - possible, like in the emacs configuration meme \"not doing work - I have to do the configuration\". Usually I do not add web-based + possible, like in the emacs configuration meme "not doing work + I have to do the configuration". Usually I do not add web-based things into emacs. ## Notes -- I like the dedicated-buffer interface (I\'m assuming using +- I like the dedicated-buffer interface (I'm assuming using magit-section and transient). - \<meain\> Very interesting ideas. I was very happy when I was able to do simple |