3 ms·

Good to see you in this thread! I just added Marginalia to the recommended search engines in a new search tool I'm building, to get programming answers faster.

by ffhhj 5y ago

Good to see you in this thread! I just added Marginalia to the recommended search engines in a new search tool I'm building, to get programming answers faster. The search assistant builds queries for specific sites with "site:targetsite.com , programming question" (that comma is not a typo). When doing a query like that I get no results but these warnings: /!\ The term "," contains characters that are not currently supported /!\ Try rephrasing the query, changing the word order or using synonyms to get different results. Tips. sample: https://search.marginalia.nu/search?query=site%3Astackoverflow.com%20%2C%20css%20background https://search.marginalia.nu/search?query=site%3Astackoverfl... Please make your engine ignore the comma, it shouldn't affect the search. Either ignore the site:... expression or filter sites accordingly. Thanks a lot for creating Marginalia!

marginalia_nu 5y ago
Ignoring comma seems doable, I'll have it fixed in a few days, currently away from my work computer. site:-queries are supported, but only at the first domain level. (e.g. site:marginalia.nu; not site:search.marginalia.nu). I might tune it so that it strips subdomains automatically, that is pretty trivial.
marginalia_nu 5y ago
I've made the changes to the syntax (as well as fixed a bug that broke site:-queries), although I think it's probably not a great search engine for searching big mainstream websites like SO, the crawler is designed to de-prioritize them over blogs and what have you. Disregard what I said in the other comment about the domain names, use the full domain name. e.g. this works fine: https://search.marginalia.nu/search?query=site%3Amemex.marginalia.nu+java&profile=yolo&js=default https://search.marginalia.nu/search?query=site%3Amemex.margi... If you just do a site:-query without any search terms, you'll get a dump outlining the crawling status of the site. Like this: https://search.marginalia.nu/search?query=site%3Astackoverflow.com https://search.marginalia.nu/search?query=site%3Astackoverfl... You can see it's crawled 80 pages but not gone much farther since SO ranked so low.

by ffhhj 5y ago

marginalia_nu 5y ago

Ignoring comma seems doable, I'll have it fixed in a few days, currently away from my work computer. site:-queries are supported, but only at the first domain level. (e.g. site:marginalia.nu; not site:search.marginalia.nu). I might tune it so that it strips subdomains automatically, that is pretty trivial.

marginalia_nu 5y ago

I've made the changes to the syntax (as well as fixed a bug that broke site:-queries), although I think it's probably not a great search engine for searching big mainstream websites like SO, the crawler is designed to de-prioritize them over blogs and what have you. Disregard what I said in the other comment about the domain names, use the full domain name. e.g. this works fine: https://search.marginalia.nu/search?query=site%3Amemex.marginalia.nu+java&profile=yolo&js=default https://search.marginalia.nu/search?query=site%3Amemex.margi... If you just do a site:-query without any search terms, you'll get a dump outlining the crawling status of the site. Like this: https://search.marginalia.nu/search?query=site%3Astackoverflow.com https://search.marginalia.nu/search?query=site%3Astackoverfl... You can see it's crawled 80 pages but not gone much farther since SO ranked so low.