@maxthelion I have a custom layout, still qwerty because I don’t have enough to time to completely start learning to type from scratch https://t.co/pIo9IsRMEf
@psychemedia That is understandable, just be careful that the tokens actually indexed by Lunr _might_ not directly match what is in the text, i.e. due to stemming etc.
@psychemedia Hey, interesting work on using Lunr with jupiter notebooks. Let me know if you run into any issues using with Lunr, I can try and help. Here is an example of doing term highlighting btw - https://t.co/QiHBse8n5S
@psychemedia Yeah, during indexing Lunr captures the position of terms within a field. By default these are not stored in the index because they increase the size of the index. It is possible to whitelist them though: https://t.co/cjB4AJ9ro0
@raymondcamden You can dump the index as JSON and then store that anywhere, so you could use IDB for example to avoid a trip across the network. To actually perform searches though the whole index needs to be in memory.
@raymondcamden Yeah, the size of the index is a big limiting factor, especially on mobile etc. I’ve been working on a more space efficient index format so at least getting the index to the client is less of an issue, so far getting ~30% reduction