#LoqiA fragmention is an experimental extension to a URL to link and cite individual words or phrases of a document using double-hash anchors (##) https://indiewebcamp.com/fragmention
lukebrooker joined the channel
#JonathanNealIt’s the idea I first heard from KevinMarks. Human readable links to individual fragments of text.
#kylewmthis betterlink thing seems to misunderstand the point of fragmentions
#KartikPrabhuno not really. Fragmention originally has double # convention. KevinMarks then added the idea to use only one # anyway since ids don't contain spaces anyway
#kylewmI'm impressed they bought a .io domain and everything for it though
#KevinMarksbetterlink replaces fragmentions with a silo o_O
#rasculafter more looking, doesn't seem like a worthwhile alternative to fragmentions at all
#JonathanNealConcern: Fragmentions are not a valid URL due to ##. Solution: Drop the second #. Concern: Fragmentions target the element and not the selection itself. Solution: Why not allow for both? html[data-fragmention-selects="*"] or html[data-fragmention-elements="p h1 h2 h3 h4 h5 h6"] html[data-fragmention-selects=":text”]. Concern: Not All Browsers Play The
#JonathanNealSame. Solution: None, if irregularly spaced will not appear as you expect in old IE≤8 anyway.
#JonathanNealSo that there could be some way to say “select the nearest ancestor” versus “select the nearest paragraph ancestor” versus “select only the text”. This could help marginalia as well, since the styles often target particular sections or paragraphs of content already.
#rasculspecifically, what could the url look like?
mdik_ joined the channel
#JonathanNeal[data-fragmention=":text”] versus html[data-fragmention=“h1, h2, h3, h4, h5, h6, p”]
lukebrooker joined the channel
#JonathanNealrascul: oh, I was saying the fragmention granularity could be defined by the page author.
#rasculahh, i guess maybe i'm not quite understanding how it could be implemented
#rasculsome of this stuff is over my head and i'm happy it just works ;)
#JonathanNeal<html data-fragmention=“h1, h2, h3, h4, h5, h6, p”> … forces fragmentions to target the nearest h1, h2, h3, h4, h5, h6, or p element that contains #some-text
#rasculso it can be defined in the html where to target?
#JonathanNealI’m open to other suggestions. The goal is to control what text or element gets selected. Is it just the text (ala CTRL+F). Is it the nearest element (ala classic fragmentions), Is it the nearest section or paragraph (ala marginalia)?
#rasculi don't have any suggestions, i'm mostly trying to understand and maybe help you figure it out :)
#rasculhrm maybe i'll finally update my ssl cert tonight heh
#rasculdoesn't look like i can hold out for letsencrypt.org without people complaining to me daily :(
#kylewmtantek: I honestly don't think that is confusion, bupkes has been around here fairly recently
#tantekeven if bupkes isn't confused, that post reads like it is conflating it all
#tantekthus it promotes (even if unintentionally) that confusion
#tantek.comedited /ind.ie (+181) "/* ind.ie is not IndieWeb / IndieWebCamp */ fix heading to not require URL escaped chars, add subsection Examples of confusion and conflation" (view diff)
#tantekthen when we have examples of replies with reply-contexts of bbs, then we can say - hey - do you want these comments in real time? webmention can do that.
#tantekthey may be more open to supporting this tech because it takes weight off their shoulders to support more signups
#kylewmit's also kind of overly humble that the bottom says "powered by digital ocean and twilio"
#tantekit felt very calming to read the recent posts stream
#tantekwith the absence of so much as any UI widgetry
#tantek!tell caseorganic "Big Boring System" seems like a good example of deliberately calm technology, e.g. check out their "recent posts" stream: https://bigboringsystem.com/discover
#kylewmlike your website, or like some crazy physics data?
#gRegor`KartikPrabhu: How "big"? I doubt your database is actually reaching a size that it would acuse problems, but indexing of columns (or lack of) can cause slow queries
#LoqigRegor` meant to say: KartikPrabhu: How "big"? I doubt your database is actually reaching a size that it would cause problems, but indexing of columns (or lack of) can cause slow queries
#LoqigRegor` meant to say: You could try a non-unique index on it, but yeah, ~200 isn't a lot of rows that I would normally expect to give that much delay.
#@jbrewer1) I miss the diverse experience of reading content on personal sites. DifÂferÂent texts deÂserve disÂtinct tyÂpogÂraÂphy/presentation. (twitter.com/_/status/582559550097600512)
#tantekit is a good counterpoint to remember, that there are numerous (majority of?) people who previously had no voice on the web, but with Twitter etc. silos they have *some* voice
#ben_thatmustbemeIf I have something tagged "Indieweb" on my site, I make it a u-category with a link to the search for all posts that have that category
#tantekben_thatmustbeme: what do you mean by "categories are actually enumerated" ?
wolftune and chalettu_ joined the channel
#ben_thatmustbemei have an ID for any category. I treat it like an object, so it can change system wide. I know, i know, plumbing
#tantekis that a cache? optimization? or do you depend on it in storage?
#tantekgenuinely curious - I don't have an approach figured out for implementing tags/categories yet myself. Was going to start with "dumb" text search backwards in time to implement the "common" use case of clicking on a hashtag and seeing most recent results for it.
#tantekyou can have multiple tags (p-category or u-category), and you *could* have multiple "u-tag-of" but that really should be reserved for POSSE copies
fahrstuhl, snarfed, torrorist, KevinMarks_, LauraJ and KevinMarks__ joined the channel
#ben_thatmustbemeI think i'm just going to have to deal with multiple copies of some data
KevinMarks_ joined the channel
#ben_thatmustbemewhich honestly makes more sense, as its stored before and after approval (though I automatically approve them for now)
#tantekben_thatmustbeme: I think that can be ok (some multiple copies) as long as you have a well defined algorithm for merging/de-duping them
#ben_thatmustbememy third normal form wired brain is saddened by this :P
dariusdunlap joined the channel
#tantekAlso important to document each *source* of such copies / duplicates, so that you can potentially fix it at the source(s) in the future.
#ben_thatmustbemethats not a property of a tag/category, but rather a tag/category as it relates to a post
#tantekdo we need to make area tags into their own objects?
#tantekagain, we should start with the display of area tags in a post, then what it is the ideal/sensible markup for it from the perspective of supporting that presentation, and *then* figure it how we can microformat it up.
benwerd joined the channel
#ben_thatmustbemewell this is about logical organization, the markup can change, but the concept that a person will be tagged in different locations within posts means that whatever that location is, cannot be stored with the person (assuming you extract out the tagged person as a category)
#ben_thatmustbemei'm not going to going to not focus on coords/shape so much, and just try to get things set up to do person tagging correctly first
#ben_thatmustbemei keep getting side tracked by other optimizations in my code
#ben_thatmustbemedropping extra tables, unifying duplicate code, etc
stream7, friedcell, tilgovi and Pierre-O joined the channel
#kylewmtantek: that would be fine with me, i can create an event this afternoon/evening
#LoqiA bookmark (or linkblog) is a post that is primarily comprised of a URL, often title text from that URL, sometimes optional text describing, tagging, or quoting from its contents https://indiewebcamp.com/bookmark
#KevinMarksso, in your theoretical world, I shoudl pull in her whole twitter bio and photo and cache it on my site as she links to a non-h-card URL from it?
#ben_thatmustbemehmm, tantek, with the change over from p-bookmark to u-bookmark and p-category to u-category, I should assume that any class on an <a> tag should be u- not p- correct?
#ben_thatmustbemerealized he has author links a <a class="p-name u-url"
#KartikPrabhuben_thatmustbeme: since you asked this is the offending query SELECT DISTINCT CAST(DATE_FORMAT(CONVERT_TZ(`bundle_note`.`pub_date`, 'UTC', 'America/Chicago'), '%Y-%m-01 00:00:00') AS DATETIME) FROM `bundle_note` WHERE `bundle_note`.`pub_date` <= '2015-03-31 01:44:30' ORDER BY 1 DESC LIMIT 1
#ben_thatmustbemethe way that query is written it has to store all the output of the query temporarily, then update them to the local TZ, THEN find only unique values within that
#KartikPrabhuhmmm that seems the right order to get the TZ conversion right
#KartikPrabhu<sigh> I am contemplating pre-rendering a version of that page and updating it on change
#ben_thatmustbemehasn't used django in some time, could you update TZ after you get the output
#KartikPrabhunot without hardcoding MySQL queries into Django
#KartikPrabhuben_thatmustbeme: like I have to learn MySQL to make it efficient, but I can change my python code much more easily
#ben_thatmustbemetantek: I would be doing the same if I were doing it with flat files. What folder do i store non-objects? how do I store webmentions? etc
#ben_thatmustbemeexact same trouble. but select * from WHERE .... and if your model is correct, changing one line of a where clause can get you any method of slicing you want, which will be faster than parsing flat files to find what you want
#ben_thatmustbemeremember Databases started as flat files with just some indexing
#ben_thatmustbememost now just contain all that in a single format to keep external changes to files from breaking indexes
#ben_thatmustbemeyou can build an index if you want in flat files, but it will be really messy and inefficient
#tantekthe difference is that at least in this iteration of flat files, we're using an open format in the flat files
#tantekwhich makes the actual storage files more portable
#LoqiFalcon is a personal publishing (tweeting, blogging, realtime syndicating) web application. There is an instance of Falcon running at tantek.com and serving/syndicating notes, articles, and occasionally other types of posts https://indiewebcamp.com/Falcon
#tantekben_thatmustbeme: re: how do you get only your posts that are articles - I read into memory one entire storage bim file at a time, and then look backwards in time through those posts until I find enough of the type of post I am looking for.
#tantekso if there a three blog posts in the most recent bim - it's one file read
#LoqiFacebook is a popular content hosting silo and activity aggregator most well known for being the largest centralized social network on the web https://indiewebcamp.com/Facebook
#kylewmtantek: if you wanted to display an /articles feed with more articles, I don't think some sort of index would be a premature optimization there
#Loqisilo quits are public statements by individuals announcing they have publicly quit posting on or using (at all) various silos, either with intended permanence, or temporarily (sometimes referred to as taking a "social media break") https://indiewebcamp.com/silo-quits
#tantekkylewm: not sure - might depend on how many articles I was displaying on /articles
#tantekkylewm: I'd say the optimization would be premature until you have measured perf > 1s display time
#kylewmwhen I was doing file storage, I added a big index for looking up posts by tag
#kylewmbut the index itself was quite large and took a non-trivial amount of time to read
#KartikPrabhuaaronpk: it seems IndieAuth is not doing the redirect properly. logging into a page redirects to wiki homepage after authorization
#tantek(aside: based on like decades of experience, optimzing for minimizing file I/O is a great place to start with perf - as that's almost always the first-order perf bottleneck for anything - I/O)