#[miklb]it’s just called Nginx Cache. Supposed to purge the fastcgi cache when content changes, or a manual button in the Tools
#GWG[miklb]: Oh, that. Did you compile cache-purge for Nginx?
#GWGIt isn't built in by default, or it used to not be.
#[miklb]I did not, but still plan to compile nginx so I can add pagespeed. I’ll make a note to look at that module as well. I’m open to suggestions for other nginx modules I should consider.
samim, leg, snarfed, eli_oat, krup and [eddie] joined the channel
#[eddie]Looks like it might be almost time to re-do my "On This Day..." code. Last Spring I added an On This Day feature that showed any posts from the previous year. This worked because I didn't have almost any posts on my site from 2016. However, 2017 was common to have near 20 posts per day... lol So looks like i'll have to add some filtering to my On This Day feature
#[eddie]I'm thinking to start, I'll make it show the first post that has a photo
#aaronpkcould maybe also rank those by number of responses
#[eddie]Yeah, my first thought is quick fix to prevent 20+ on this day posts :) Second thought is after that is done, then an algorithim
#[eddie]The three things I think I want to rank it by is: Does photo exist, Is anyone tagged and are there responses?
#[eddie]I'm not sure of the priority order of those
#[eddie]Yeah, and third maybe using post type discovery to give certain post types a higher priority. For example: An article I posted with no one tagged should probably be higher than a movie I watched with my wife
#[eddie]The movie would have a person tag, so that could potentially end up higher without post type discovery weighting
#[eddie]I guess start simple with person tags and responses with a boost if it contains a photo
#[eddie]Then when I come across concrete days where I feel like that wasn't the best post to feature, I can begin tweaking the algrothim
#Loqiincremental has 1 karma in this channel (6 overall)
#aaronpk[eddie]: I was thinking about somehow getting weather info into my main Compass location database, that way the weather info is always accessible to anything that has access to my location. it would simplify my micropub endpoint code a lot to have that weather info already available. is that something you'd find useful as well?
#[eddie]aaronpk: Yeah, that would definitely be useful. I store quite a bit of data from dark sky in regards to weather but most of it is just so I always have it if needed. The main things that I currently use or have active plans on using are: apparentTemperature, icon, moonPhase, nearestStormDistance, precipType, summary, sunriseTime, sunsetTime, temperature, visibility,
#aaronpkI was thinking about moving the code there
#aaronpkto have the app look up weather info and include it in the location updates it reports to the server
#[eddie]Yeah, that seems like that could work. I think the only concern is you can have GPS location without internet
#aaronpkyeah, i'm on the fence about doing it on the phone or doing it in Compass
#[eddie]So if the phone losses internet, do you lose weather tracking or does it just collect all the non-weather locations when you get internet again?
#aaronpkwell before we get that far, I was also thinking about how to handle caching/rate limiting. the weather APIs have pretty low limits of how often you can ping them, certainly much less than looking up every record
#aaronpkso I was thinking about caching the weather info based on both time and distance
#[eddie]That's true. I don't have to worry about limits currently because I only do it per post, and I don't do that many daily posts yet haha
#aaronpkbut for a several-hour plane flight, it would stop recording weather
#aaronpktho at that point, the weather isn't very relevant anyway
#[eddie]True. I guess the only other concern is battery life
#aaronpkyeah, network fetches are a big battery hog for sure
#[eddie]Yeah that might make be a bit nervous for my phone. I try to be somewhat conservative with my GPS tracking. (I have it pause when I'm at locations for an extended period of time)
#aaronpkthe caching i'm imagining would be time *or* distance, so it would try to get new weather every 5 minutes, but also if you moved more than 5km in 5 minutes
#[eddie]hmm yeah, right now I have the send interval set to eery 30 minutes to reduce network requests
#aaronpkso the other option I was thinking was having Compass augment the location records it receives with the weather info
#aaronpkthat's possible with the darksky API since it lets you do historical lookups
#aaronpkbut, a few challenges I was running into with that approach:
#aaronpksince Compass writes to append-only files, it can't really go back and update records that are already written to disk. this means it has to fetch the weather info before it writes to disk
#aaronpksince the app may send very large batches of locations that require multiple weather lookups, it obviously can't do the weather lookup in the same HTTP request that the phone sends
#aaronpkso this implies Compass needs to take the locations from the phone and write them to a temporary location, probably a Redis list to make things easy
#aaronpkthen a background script can go through and process those locations to add the weather info, then write to the storage file
#[eddie]Hmmm yeah. That is tricky. As I was reading though that, some kind of queue came to mind, aka Redis list. So that definitely makes sense, but I definitely see how that can be a lot of added complexity to the system
#aaronpkbut again since Compass writes to append-only files, this process can't happen in parallel like I'd normally do, it has to be done with only a single background process in order to avoid conflicts writing to the file
#ZegnatThat feels like a storage nightmare. Would it not be easier to at that point have a separate storage for the weather data?
#sknebelsecond quartzdb as a queue that gets deleted at end of day instead of adding another dep?
#[eddie]ahhh yeah. lol I was just about to say, you probably can't do parallel with your database
#aaronpkZegnat: I already have a separate DB for the weather data, but right now the problem is keeping the two relatively in sync and it makes my micropub code a lot more complicated to pull the info from two places
#[eddie]Hmm Zegnat has an interesting thought there. What if there were two quartzdb that compass talked to
#[eddie]so one API but two quartzdb that get merged in Compass?
#ZegnatAh, I guess you have it all hooked to your Micropub, yeah. I was thinking only of Compass as a server.
#aaronpkmy micropub endpoint wants to be able to either query the last location (with weather) or query the location and weather at a given timestamp
#[eddie]Does this sound do-able at all? Compass receives data from Overland, it writes location data to location.db. Periodic background task in Compass fetches the last entry from weather.db, grabs all locations from location.db that are more recent. Fetches weather for those locations and appends them to weather.db
#[eddie]Then when you query Compass's API (either last location or given timestamp), it fetches the location.db and weather.db for given timestamp
#aaronpkI guess if the timestamps matched up exactly that could work
#[eddie]If you used Dark Sky's API, you could use the timestamp from the location data when querying it, so then you could ensure it had the same timestamp
#aaronpkanother thought: Compass could use MySQL as the queue and temporary storage, since I can write to MySQL in parallel. so Compass receives a batch of locations and writes it to a MySQL table, then hangs up the HTTP connection and kicks off a background process. That starts querying the weather API for each location (with the 5min/5km buffer) and writes either the successful weather info or a failure code
#aaronpkto the MySQL record. when that batch is done, it's written to the file on disk.
#ZegnatI don’t have a stake in this as I am on Android, but I would say whatever you come up with that saves the phone from doing the network requests is the right move
#ZegnatAlthough constant GPS tracking is definitely something I am interested in doing. The timed GPS polling my phone does now doesn’t really give me a lot of data.
#[eddie]Yeah timed GPS is tough. Misses some nuances
#aaronpkre: long pauses at locations, iOS actually provides that data pretty well now
#ZegnatIt also seems that if the GPS chip isn’t constantly running and looking for signals, it is simply less accurate. When it gets turned on and has limited time to figure out where you are before being shut off again it isn’t going to be as accurate.
#aaronpkif you have "include tracking stats" enabled in Overland, you'll see some records that are labeled "visits"
#ZegnatAt least that is how it seems to me, comparing my GPS logs with, e.g. just keeping a maps app open for a bit
#aaronpkZegnat: yeah that usually has to do with the type of location the app requests. that's why I have all the little toggles in Overland, so you can customize it
tantek joined the channel
#[eddie]ohh really? I'll have to turn on tracking stats and test that out
#aaronpka "visit" is recorded about 10 minutes after you stop moving. it's surprisingly good
#aaronpkZegnat: yeah, iOS calls it "navigation", "fitness", etc, as well as lets the app request location at a certain accuracy
#aaronpkandroid has similar features IIRC but i'm not as familiar with all the recent updates there
#ZegnatMe neither. The app I use lets me set minimum GPS accuracy to a minimum of 4 metres.
#aaronpk[eddie]: I was actually thinking about making a separate tracking mode in Overland that doesn't request location updates at all and only uses the significant change and visit APIs. it would basicaly use no battery at all.
#aaronpki've hesitated on implementing that because I know I won't use it
#[eddie]aaronpk: That's interesting. I definitely want more in-depth tracking than that as well. That's why I fiddle with mine.
#[eddie]When I leave my house, I typically disable the Pause Updates Automatically
barpthewire joined the channel
#[eddie]and I'll enable Trip Tracking and have best accuracy and such.
#ZegnatNot sure why Overland users would want to not use constant tracking, aaronpk? Wouldn’t there be tonnes of apps that can offer that?
#[eddie]I just am at home a lot so I try to not have it run when I'm at home all day
#aaronpkZegnat: there are very few apps that let you point the app at your own server
#aaronpkI know some people are using the app to get location data on people for studies, and that often involves needing to know just rough location liek whether they're at work or at home
#ZegnatI do wish Backitude let me switch into some sort of “trip tracking” mode. To switch from GPS polling to full on tracking when on a bike etc.
#tantekaaronpk, then keep using @-nickname for user-entry, but just for lookup in your nicknames cache and then display their actual name, linked to their site
#tantekin a post when you reference someone you really should link directly to them, not to a local page about them, that also seems like the kind of thing a silo or SEO-skeezy site would do
#aaronpkwho said I was going to link to my local copy of them in a post
#ZegnatI don’t think aaronpk is actually going to link to those h-card pages?
#ZegnatI meant he is generating URLs because the h-cards in his system have a URL. Not that he would explicitly refer to those h-cards. Sorry for confusion.
[cleverdevil] joined the channel
#ZegnatPresumably those URLs are only for him. Thus I think it is fine to use @nicknamechosenbyaaronhimself in the URL.
#tantekI like the Wikipedia (and frankly journalistic approach) in articles, first reference is full name ("name") linked to their site, subsequent references can use just given name if informal (a friend) or family name if a more formal reference (someone you don't know and you're citing them)
#aaronpkbut that's just not how peopel write in microblog posts
#aaronpki'm not going to try to apply journalistic writing techniques to microblogging :P
#tantekthat may be a reason to distinguish, depending on length? like if a microblog post gets long enough with enough paragraphs it starts to feel more like an article
#tantekhow do @-names in your posts work when you POSSE to different destinations? do you have to keep a list of translations? what about passive POSSE e.g. micro.blog where it just reads your feed?
#tantekeither you have to deal with @-name clashes across services like that, or you come up with something else (which is why I like @-domain as a indieweb-aware way of doing it)
#aaronpkI only recently got the nicknames cache working again so I still need to do the mapping to silo accounts but the data is in place
#tantekit's also so much better than the @-@- thing of mastodon etc.
#aaronpkI only have one h-card right now, sknebel's
#aaronpkAnd that's cause he doesnt have that twitter account
#[eddie]tantek: on my website @nickname is a URL that points to their homepage. That same href is displayed on Micro.blog
#[eddie]I am pretty sure micro.blog doesn't care about the @name as much as the URL it points to
#[eddie]obviously for micro.blog it would need to point to their micro.blog profile instead of homepage
#[eddie]but manton hasn't finished adding that part yet
#ZegnatI would love to just see what names people end up using for me
#[eddie]If someone is at least a casual acquaintance, I typically use @Firstname unless there is a first name conflict, then I add @FirstnameLastname or some other identifier like I use my niece's @FirstnameMiddleName
#[eddie]unless someone has a well known usename like aaronpk
#Loqischmarty: [eddie] left you a message 4 days, 16 hours ago: While parsing https://martymcgui.re/2017/12/26/102358/ I realized that your dt-published date is ISO8601, but the dt-start is Jekyll’s offshoot date where you place the “T” with a space and add a space between the time and the timezone
#Loqiok, I added "http://www.ftwynn.com/articles/creating-a-post-api-for-hugo/" to the "See Also" section of /static_site
#loqi.meedited /static_site (+63) "schmarty added "http://www.ftwynn.com/articles/creating-a-post-api-for-hugo/" to "See Also"" (view diff)
#sknebel[keithjgrant]: Telegraph only looks in the first mf2 object on the page it seems
#schmarty!tell [eddie] thanks for the dt-published vs dt-start inconsistency on my post. i'm probably forgetting to pass in a format somewhere.
#LoqiIt looks like we don't have a page for "prefetch" yet. Would you like to create it? (Or just say "prefetch is ____", a sentence describing the term)
#[keithjgrant]Is it valid the way I've got it now? H-card followed by h-entry?
#LoqiIt looks like we don't have a page for "rel-prefetch" yet. Would you like to create it? (Or just say "rel-prefetch is ____", a sentence describing the term)
#gRegorLoveNoticed [keithjgrant] has that, in case wondering where ^ came from
#ludovicchabanthello there! happy new year! are there any good tools I can use for POSSEing stuff to Mastodon/GNUSocial?
[eddie] joined the channel
#[eddie]keithjgrant Wow Beautiful Site redesign! How did you do the page transitions?
krup joined the channel
#KartikPrabhu [eddie]: [keithjgrant] seems ti be using Javscript to do page transitions, afaik that is the only way to do them
#[keithjgrant]eddie Thanks! I listen for clicks on links; if it's to a same-domain url, I prevent navigation and load the page via AJAX instead. All pages have a few key classnames ("js-main" or something like that) to help find the key part of content to transition in.
#[keithjgrant]I do a lookup based on old url & new url to decide which animation to play. Used GSAP for the actual animation implementation
#[eddie]ohhhhh nice! So you have static site pages, and you're just pulling in the content via JavaScript and injecting it
#[eddie]That's pretty awesome. I'm stuck in-between static sites and doing a server-side Angular. So this is an interesting medium
#[keithjgrant]My one big compromise is I couldn't put an `h-entry` on the <body>, since some pages need it and some don't. So I had to nest that more deeply, but I think it works out okay this way
#ludovicchabantAh thanks Zegnat ! But AFAICT Bridgy is generally for the other way (i.e. getting pings back from somewhere onto your website's IndieWeb endpoints)
#ludovicchabantoh Bridgy does both ways? Mmmh I must have missed that
#ludovicchabantEspecially since the second paragraph says "This isn't syndication or POSSE!" :D
#ludovicchabantBut then later, "Mastodon: replies, likes, and reposts aka boosts, both directions, via ActivityPub".... I guess "both directions" is the key here
#ludovicchabantok so if I understand this correctly, both Bridgy versions will crosspost on various services if you send them a webmention to tell them to crosspost a specific URL, is that correct?
#ludovicchabantso I would need to integrate something in my CMS that figures out when there's a new article, and then ping Bridgy about it?
#snarfedludovicchabant: roughly. bridgy, yes. bridgy fed, only responses, not original posts. (and it federates, not cross posts.)
#snarfedand yes, but bridgy also has an interactive UI, so you don't have to do webmention
#ludovicchabantwhat's the difference between federating and crossposting?
#Loqifederation in the context of the indieweb refers to services and features on indieweb sites that work directly with other indieweb sites peer-to-peer, without having to setup or create accounts in both places, and without being bottlenecked by any kind of centralized service or silo https://indieweb.org/federation
#LoqiIt looks like we don't have a page for "crossposting" yet. Would you like to create it? (Or just say "crossposting is ____", a sentence describing the term)
#snarfedfederating with e.g. mastodon means it understands that the real version of your post is on your site, not on mastodon
#ludovicchabantso cross-posting means "copy the whole thing" (assuming it's under 280 chars or whatever the limit is) and make it look as if you posted it natively on Twitter/Mastodon/etc?
#ludovicchabantwhereas federating means just posting the title and link?
#LoqiCross-posting is the social media practice of syndicating or placing the same content, possibly slightly modified, on numerous different websites https://indieweb.org/crossposting
#ludovicchabantoh right Zegnat -- yeah it's especially confusing since I don't know if "federation" for indieweb has the same meaning as for Mastodon/GNUSocial/etc. :)
#snarfedludovicchabant: federating doesn't necessarily change how the content of a post shows up. it has to do with how the underlying sites interact. kinda subtle.
#snarfedindieweb protocols are federated, yes. each site is a node
#snarfedbut via different protocols than mastodon (activitypub) or gnu social (ostatus)
#ludovicchabantah ok, so yeah in my case I already have a Mastodon instance myself... so my indieweb site "identity" is different from my Mastodon "identity"
#ZegnatAnother interesting question is if you need a separate Mastodon identity if your posts on your website all are compatible with other Mastodon instances already 😉
#Loqi[raymestalez] #1441 What is necessary for Mastodon to be able to fetch my profile and a list of posts from my blog?
#ZegnatAh, O thought that was more of a "solved" case then it appears. My apologies.
#ZegnatHas anyone tried using Mastodon as their IndieWeb CMS?
bengo joined the channel
#snarfedZegnat: you can definitely follow a web site with an atom feed from mastodon. it's finicky, but doable
#ludovicchabantZegnat: my use case is that i want to treat Mastodon the same as, say, Twitter... which is that the fact that I own my Mastodon instance is irrelevant and just a “coincidence”
#ludovicchabantThe reality is that after thinking about it for a while, I don’t think I want to centralize as much of my content on my website as so if you do... I want to cross post, but then I want to “engage” with other services’ users directly on the services themselves, rather than from the confines of my indie web site
#snarfedludovicchabant: you can still do that! eg i generally POSSE everywhere manually, and often like/reply within those services (and automatically PESOS those back)
#ludovicchabantYep I can but I guess for instance Fed Bridgy doesn’t fit it? It would have to know about my Mastodon instance to cross post?
tantek joined the channel
#ludovicchabantThere’s also the fact that I’m using a static website so unlike, say, Wordpress, I don’t have the concept of “publishing an article” in the same sense
#snarfedif you really want to POSSE to mastodon, then no, i don't think any current indieweb tool does that, including bridgy
#snarfed(i think static vs dynamic is unrelated to this)
#snarfedtry just POSSEing to mastodon manually for a while
#ludovicchabantI do think that static/dynamic has a bit to do with it since some services like Bridgy support automation based on some CMS sending a POST request on publish
#snarfedsure, via webmention. there are lots of tools that help static sites send webmentions.
#tantek.comedited /IndieAuth (+237) "/* how to should include endpoint definition */ stop suggestion extra steps for the 99% that are only needed for the 1%" (view diff)
#tantek.comedited /IndieAuth (+199) "stub Setup your own IndieAuth provider with ask someone who has done it" (view diff)
#tantek.comedited /IndieMark (+159) "/* draft levels */ note setting up IndieAuth can be done easy with rel-me, or harder with local provider support (no obv how to on [[IndieAuth]] to link to, so yeah, much harder)" (view diff)
#tantekKartikPrabhu: re: Known pivoting discussion, perhaps #withknown ?
#KartikPrabhutantek: sure, I keep forgetting about all the channels :P
#LoqiIt looks like we don't have a page for "channels" yet. Would you like to create it? (Or just say "channels is ____", a sentence describing the term)
#tantek.comedited /discuss (+575) "/* Chat Channels */ add brief descriptions for related channels used by the community and available in Slack #microformats #knownchat #bridgy" (view diff)
#loqi.mecreated /LS (+20) "prompted by tantek and redirect added by tantek" (view diff)
#tantekKartikPrabhu: ^^^ added explicit #knownchat docs to make it a bit more discoverable hopefully
#LoqiIt looks like we don't have a page for "minimum" yet. Would you like to create it? (Or just say "minimum is ____", a sentence describing the term)