#dev 2019-02-04
2019-02-04 UTC
# gRegorLove js;dr? :)
KartikPrabhu and [tantek] joined the channel; KartikPrabhu left the channel
# Loqi It looks like we don't have a page for "hydration" yet. Would you like to create it? (Or just say "hydration is ____", a sentence describing the term)
[eddie], [Serena], snarfed, KartikPrabhu, tglobe, cweiske, swentel, jeremych_, iasai, [jgmac1106], swentie and barpthewire joined the channel
# petermolnar I'm still tempted to do something with my scrobbles, but the mere csv file of timestamp,artist,album,track,artistid,trackid,imgurl is 10MB, so it's not really an option to spit that thing into a single HTML page
iasai joined the channel
# petermolnar oh, wait. only 5MB after deduplication
# petermolnar add all the markup for rdfa, mf2, plus the json+ld invisible content version, é violá, I present you a ~30MB HTML table
[kevinmarks] joined the channel
# [kevinmarks] extending xoxo to tables was something I thought about but didn't have a personal use case for
[metbril] and iasai joined the channel
# petermolnar no, but it's a still a shithole: eg. google structured data testing tool can't parse nested <a href="mailto:xyz@domain.com" property="email">xyz@domain.com</a> properly
# petermolnar which worked perfectly fine with microdata
# petermolnar and if there are different implementations to read a standard, it's useless
# petermolnar I'm well aware I shouldn't "design" my markup for Google, but the reality is that's the only thing that might consume it, so I will make it Google-compat
# petermolnar I'm not going to add invisible json+ld as well, but I have considered adding a side-file .json and purge the rdfa thing from my site
# petermolnar with an alternate text/json meta pointing at it
# petermolnar however, building my category pages in json is a pain
# petermolnar a serious pain
# petermolnar that I don't know, I found some github links yesterday which suggests it should
iasai and [schmarty] joined the channel
# [schmarty] GWG: publishing it is one thing and making sure the ecosystem of parsers and consumers supports it is another dozen things.
iasai, eduardm, [kevinmarks], eli_oat, [Rose], [eddie], tantek, deathrow1, jackjamieson and [cleverdevil] joined the channel
# [cleverdevil] So, thinking more about my work on nicknames and my "pseudonym" prototype.
# [cleverdevil] Two things I've noticed:
# [cleverdevil] 1. I made a false assumption that all core identities will be *domains*. I think this is wrong. Some people will actually have a core identity at a full URL. I am specifically thinking of @Zegnat, whose identity isn't at the root of his domain. I'll need to make a shift here, I think.
# [cleverdevil] 2. Some people have multiple identities with individual silos. I made an assumption that any given "domain" would have a single "identity" on a target. Not true. I'll need to shift this as well.
# [cleverdevil] I'm going to make a shift on both of these when I get a chance, hopefully sometime this week, and also start indexing additional information and improve searching, specifically by h-card nicknames and by usernames in individual silos.
# [cleverdevil] I am still not sure if a global service like this is the right answer, but its worth continuing to see where this rabbit hole leads 🙂
[tantek] joined the channel
# aaronpk how are you populating the data? is it from /chat-names or the webring or something?
# [cleverdevil] Its done on-demand.
# [cleverdevil] As a domain is requested, I fetch it, cache it, and then it becomes searchable.
# [cleverdevil] Once a domain is requested again, I pull it from the cache if its been indexed within the last 24 hours, otherwise, I re-fetch and update the cache.
# [cleverdevil] Correct, that was the idea.
# [cleverdevil] Yeah, I think that's about right, if I understand xray.
# [cleverdevil] Yes.
# [cleverdevil] Plus searching and content replacement.
# [cleverdevil] I *could* leave the content replacement to the CMS, but its super easy to do in the service, and it makes it easier for people to integrate with their CMS, so I figured "why not."
# [cleverdevil] One challenge with the content replacement is the case where a mentioned identity has multiple identities in a single silo.
# [cleverdevil] I could pick one, or return both.
# [cleverdevil] If picking one, which one do I pick?
# [cleverdevil] If I return both, I am really just passing on the responsibility of that decision to the caller, which isn't really very helpful to them 😉
# [cleverdevil] (This is an edge case... the vast majority of people don't have this problem... [aaronpk] was the identity that I found that made me think about this).
# [cleverdevil] ^^
[schmarty] joined the channel
# [schmarty] [cleverdevil] have you encountered that case in the wild yet?
# [schmarty] lol chat timing
[jgmac1106] joined the channel
# [cleverdevil] Ah, good idea.
# [cleverdevil] Well, that solves one part.
# [cleverdevil] I also ended up using MongoDB, because I wanted to tinker with the new AWS MongoDB service.
# [cleverdevil] Which is good... but way too expensive for something like this 😛
KartikPrabhu joined the channel
# [jgmac1106] Yeah it would have been super easy I could reply to unlimited posts in Known
# [jgmac1106] Love to hear your thoughts [cleverdevil] I pay for Atlas but that is more for way to support open source database developers plus use their 15 second backup
tantek_ joined the channel
# [cleverdevil] Oh, on DocumentDB?
# [cleverdevil] Its great, but is an older version of MongoDB because of licensing.
# [cleverdevil] Its value over Atlas is that its fully integrated with the rest of AWS.
# [cleverdevil] So you can use all of the other AWS "native" goodies like CloudFormation, CloudWatch, etc. with it.
# [cleverdevil] I suppose I *could* use Atlas' free tier for this use case...
snarfed and iasai joined the channel
# [jgmac1106] Where is Zegnat?
# aaronpk it does, as soon as it can index your site https://pseudonym.cleverdevil.io/search/aaronpk
# [cleverdevil] Yes, I noticed your note 🙂
# [cleverdevil] Definitely will fix both issues - full URL identities and improved searching by nickname and silo identity
leg, jackjamieson and [grantcodes] joined the channel
# [grantcodes] With this people cache stuff I always kind of imagined there being a centralized service that would already be filled with a bunch of data and handle parsing mf2 sites and all that for you but you would also want to extend it in some way with your own personal friends etc. Maybe that involves making your own service that fetches from the centralized service or there is a private customization part built into a centralized service I didn't t
jgmac1106 joined the channel
# [grantcodes] Yep, been trying to keep up, but not sure I've seen anything about centralized + customized
snarfed joined the channel
# [grantcodes] Aha, thanks people. Missed that part
[schmarty], snarfed, [eddie], gRegorLove_, jackjamieson, [tantek], [adam], [jgmac1106], koddsson and jgmac1106 joined the channel
eli_oat joined the channel