Nostr View

mleku @mleku - 5d

oof, hard day at the fiat mine today, first was the heroku and all that stuff, and then second session i discovered that some old tests in the code were not calculating time offsets properly and giving me useless numbers the new code is working fine tomorrow is code cleanup time... gotta bring everything up to date... stuff that evolves like this often has a way of getting crufty and needing a polish, but i'm just a bit in a tizz because i was pretty sure it was all working correctly, and now i see there is a lot of duplicaiton in the code so ... yeah... old code is gonna get removed, and the tests replaced with the current version and then i suppose it will be debugging time timezone data is a frickin nightmare, i'm probably going to have to thoroughly revise my timezone data code as well, it seems to me like what i really should have is a self-regenerating dataset to use, hell, it will go in the database too, because why not, then it will have an automatic update from all the sources and hopefully none of them break or anything before some tinpot country changes its time zone rules again and my code is not working liek it should be fucking governments, get in your way, even when you avoid dealing with them as much as possible

1
0
3

mleku @mleku - 5d

there was one bright point in the day in the early part of the afternoon session... dug up an old fuzzy matching algorithm from @mattn well, i didn't get a thingy on his nick to refer to him, but he's an old hand at this Go stuff, the repo was 7 years old was very nice, i've adapted it to create a fuzzy match value on different phrases (team names and achievement names) that gives a fairly good measure of the similarity going to be using this algorithm later also, it will let me get more precise matches to refine the reverse index full text search, after it finds the whole word matches in the content field, and sorts them by this metric, it can also then optionally use this to further sort similar levels of match to favor phrase matches without the full expense of phrase matching it also could prove useful with #alexandria to search a specific document (tree structured collection of document segments) to improve word searching in that it will find similar phrases, like for example with an extra word or a missing word or one or two words different in a 5 word phrase

0
0
3
Showing page 1 of 1 pages