#dev 2018-12-28

2018-12-28 UTC
iasai, eli_oat1, snarfed, snarfed1, Asquith, strugee, [tantek] and [relapse] joined the channel
#
@hafelg Webmentions: Enabling Better Communication on the Internet. #DigitalTransformation #UX #Dev #Web #Standards #Innovation #StillRelevant via ⁦@alistapart⁩ and ⁦@ChrisAldrich⁩ https://alistapart.com/article/webmentions-enabling-better-communication-on-the-internet (twitter.com/_/status/1078516027816849408)
iasai, strugee and snarfed joined the channel
#
pstuifzand !tell petermolnar I released a new version of ek (0.8.4), it fixes a problem it had with some token_endpoints and adds more error handling. Have you tried it?
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi pstuifzand: petermolnar left you a message 10 hours, 26 minutes ago: got ek working against aperture, seems like the problem is with my eksterd
iasai, ichoquo0Aigh9ie, snarfed, krychu, swentel, jjuran and [Vincent] joined the channel
#
[Vincent] !tell [jdpinto1] did you manage to fix your problem with editing the frontmatter?
#
Loqi Ok, I'll tell them that when I see them next
iasai joined the channel
#
@janboddez ↩️ Done. Added outgoing webmentions, too, after having discovered https://github.com/indieweb/mention-client-php. (Not sure if I should keep the functionality, though, or move it to another, much smaller plugin.) (twitter.com/_/status/1078594929667043330)
cweiske, mlncn, swentel and [jdpinto1] joined the channel
#
[jdpinto1] [Vincent] Not quite. I've found a temporary workaround to removing unwanted frontmatter using `formatted.content = formatted.content.replace(/slug: .*\n/, '')`, but this is clearly less than ideal. I simply can't figure out what document/code it's using to format the final file that gets pushed to my GitHub repo. I figured there was some sort of Jekyll template it was using and that I could easily modify that, but I'm really at a loss. Even s
#
[jdpinto1] specific words such as `slug` or `category` provides no clues.
#
[jdpinto1] The main modifications I want to make are to only include frontmatter variables that I want and also to modify the slug so that it's made up of the first three words in my post, not just a random number.
mlncn8 joined the channel
#
[Vincent] I’ve had a little dig through the code. I think you can specify the link style in the config file perhaps. https://github.com/voxpelli/webpage-micropub-to-github/blob/master/lib/config.js
#
[Vincent] permalinkstyle var?
iasai joined the channel
#
sknebel [jdpinto1]: the frontmatter is generated here: https://github.com/voxpelli/node-format-microformat/blob/master/index.js
[jgmac1106] joined the channel
#
@bradenslen Bookmark: Displaying Webmentions with Posts · Amit Gawande https://ramblinggit.com/2018/12/bookmark-displaying-webmentions-with-posts-%c2%b7-amit-gawande/ (twitter.com/_/status/1078614865407807489)
jgmac1106, iasai and gkbrk joined the channel
#
GWG Morning all
barpthewire and iasai joined the channel
#
sknebel good morning
#
sknebel http://openpoimap.org/ <- this shows POIs from openstreetmap - could be interesting to judge if it's worth exploring using OSM more
#
Zegnat I wonder what “BBQ” is as an amenity type. Gothenburg has a couple of those listed
#
jjuran Grill available for use?
iasai and [jgmac1106] joined the channel
#
sknebel Yes
#
sknebel "A permanently built place for having a BBQ"
#
sknebel (although potentially someone might have also tagged an explicit "BBQ allowed here" corner of a park etc as such too, even if not the intended use of the tag)
#
Zegnat Interesting
#
Zegnat So those may be more common around where I live than around Gothenburg
#
Zegnat Will need to have a look
#
[jgmac1106] of course a BBQ is its own entity
#
[jgmac1106] Then you could have sub category, Louisiana, Kansas City, Memphis
#
[jgmac1106] Ohh a grill to use.. Not a type of restaurant..
#
[jgmac1106] Yeah each campsite in US uses special icon if grill available
#
skippy i have a vague, poorly thought-out question about options for displaying notes on my own site. I have almsot 12000 tweets currently archived on my site. Each is its own file. That now seems extraneous. Im thinking of putting them in one file per year.
#
skippy that would be closer to 1K items per file. Then use HTML anchors to create permalinks for each one.
#
skippy Curious to get an idea from this group what your reactions to such a setup might be. Bad user experience? Good idea? Dont care?
#
[jgmac1106] Following with interest as I avoided this not to add 30k html files to my site
#
skippy i know that obvious answer is "do what you want, its your site" but im wondering what initial reactions one might have to loading something like /2008/#06130801 inside a page with a thousand items...
#
skippy On the one hand, who really cares, as long as each item has a permalink.
#
skippy on the other hand, would this complicate reply-to functionality for people in unexepcected ways?
#
Zegnat It will probably/possibly complicate parsing with current tools as those often haven’t encountered fragment-permalinks
#
Zegnat I personally like it though
#
Zegnat Actually, I think if you go to tantek’s really old blogs he was doing exactly that
#
Zegnat On functional side: at some point you probably need to worry about the HTML file size and the DOM complexity. As that can slow down things for actual people visiting the URL.
#
Zegnat But as long as you feel those things are within what you are comfortable with offering? I see no issues
#
[jgmac1106] @zegnat does overall website size effect functionality in the same way as page size? would there be an advantage to not publish each note as a url? but a fragment?
#
skippy file size would be big, but it would mostly be plain html. i have a minimal css, and not doing anything internally with JS
#
[jgmac1106] I have 39,000 notes to import]
#
skippy then the technical implementation comes down to storing all raw tweets in YAML or JON for use with Hugo.
#
Zegnat So the problem I am thinking of is that, if the file is big and the browser streams the HTML file and starts rendering partial DOM, or just starts rendering partial DOM because doing the whole DOM tree at once is too much, the visitor will not jump to the right place in file immediately.
#
Zegnat It could happen they just see the start of the file, thus reading entirely the wrong notes.
#
Zegnat And if they start scrolling, I don’t remember if browsers will hijack that and send them to the right place or never do that
#
[jgmac1106] so then maybe go to month or week instead of year @skippy
#
Zegnat That’s where I see the functional – or maybe usability – aspect play up. Technology wise it really doesn’t matter how you break things up
#
skippy interesting. is there common guidance on how big is too big for a singular html page? (knowing full well that the number of variables in play makes a baseline pretty well impossible0
iasai joined the channel
#
[jgmac1106] 200-300k was what I was always told..but this is a community that strives to see who can get closest to 1k
#
skippy i have a 6 line CSS inline in the html. its super minimal.
#
KartikPrabhu you could also do a page per month
#
skippy yes.
#
KartikPrabhu or week depending on the frequency
#
Zegnat I guess if you want to minimise number of HTML files, the question is just “how much data is too much data”
#
skippy sure
#
skippy i need to do some experimenting, to see how big year vs monthly archives are; and also balance that against how long it takes to geenrate the site with each.
#
skippy thanks for the discussion, folks! good stuff
#
sknebel (for reference, cleverdevil at some point posted a 1 MB feed with 3973 entries, which I've been using for some performance work - parsing that takes around a second pure parsing time)
#
sknebel (in mf2py)
#
sknebel regarding the POI stuff: I wonder if it'd be realistic to make a POI "API" that offloads some of the work to the client and just is a lot of pre-processed static files hosted somewhere
#
KartikPrabhu what is POI?
#
Loqi It looks like we don't have a page for "POI" yet. Would you like to create it? (Or just say "POI is ____", a sentence describing the term)
#
sknebel POI is Point of Interest, a location where something is, e.g. [[venue]]s for [[checkin]]s.
#
Loqi ok
#
sknebel e.g. if you check the first time for venues in e.g. central Gothenburg, your server (or your mp client) pulls a file related to that location and caches it, and can answer queries from that area from it.
#
sknebel so the individual user doesn'T need a global POI database, and the global POI database has little dynamic work for answering queries and thus can hopefully be cheap
#
Zegnat Basically open-source Foursquare, is the easiest way to describe that API?
#
sknebel yes
#
sknebel and with as little as possible complex server-side stuff
#
sknebel (although I'm mostly going by rough reports on how difficult the dynamic part is - would be interesting to try how well that works if tried really cheaply too)
eli_oat1, iasai, jjuran and benwerd joined the channel
#
petermolnar [jgmac1106] re resume vs schema.org: I used a top level Person, worksFor as previous workplaces with Org/Role type, alumniOf for my education place, knowsAbout on the Person level for skills. Mostly works. Ugly as hell.
#
Loqi petermolnar: pstuifzand left you a message 10 hours, 45 minutes ago: I released a new version of ek (0.8.4), it fixes a problem it had with some token_endpoints and adds more error handling. Have you tried it?
#
petermolnar KartikPrabhu: I knew something was missing! I should have grabbed our fax server a couple of months ago when it was decommissioned :D
sebsel joined the channel
#
petermolnar !tell pstuifzand I got ek working with aperture, it's completely happy now; didn't have time to poke at eksterd yet
#
Loqi Ok, I'll tell them that when I see them next
#
petermolnar back on schema vs resume: it's not just ugly, it's also undocumented, unprecedented, etc., even though it's sort of a core feature for online data in my opinion
#
petermolnar so what I did was completely out of my head, trying to make sense of the schema.org mess
#
petermolnar (because I was bored at that point in time)
[kevinmarks] joined the channel
#
[kevinmarks] Month works well for twitter archives, and thta is how they deliver them too - kevinmarks.com/tweets
jjuran, iasai, eli_oat1 and [tantek] joined the channel
#
[tantek] re: big HTML file and fragments for permalinks, yes as Zegnat pointed, my original blog started that way: tantek.com/log/2002/08.html - one file per Gregorian month
[jdpinto1] joined the channel
#
[jdpinto1] [sknebel] This is definitely getting me closer to an answer. Still, I have to ask what must be really newbish questions. I can’t figure out how voxpelli’s webpage-micropub-to-github is using his node-format-microformat code. I’m guessing it’s somehow referencing the repo on GitHub directly? If so, should I fork node-format-microformat, make my modifications there, and then somehow link my forked webpage-micropub-to-github to it? I fee
#
[jdpinto1] over my head with this. All I want at this point is to change the slug so that it uses the first 3 words of my content instead of a random number.
#
Zegnat tantek, That was an inspiration for https://grapefruit.zegnat.net/ as well, which was my test for blog-as-monthly-html-files
#
[tantek] back then (2002) I tried to use Yahoo Groups as a way to create longerlasting permalinks for what were news articles that typically disappeared after a week or so, then it turns out Yahoo did some sort of purge of Yahoo Groups a while ago, perhaps due to inactivity so those links died too. but at least they actually go to a page saying the group doesn't exist
#
[kevinmarks] monthly pages with id links for posts was the common pattern in early blogging
#
Zegnat I still like seeing those pages
#
[tantek] They did seem to load fast enough (even back then!) to scroll quickly to the right spot, rather than the load & jump
#
[kevinmarks] eg http://epeus.blogspot.com/2002/11/
[jgmac1106] joined the channel
#
[jgmac1106] @zegnat @petermolnar yes I was thinking the edu vocab, for my use case, it is highly and overly devolped a large vocabulary controlled by one membership org and an open source movement that kinda tries to tag along
#
Zegnat I have seen the load and jump a couple times these days. But that is on 3G-ish connection with what goes for modern design these days. It is very possible that even some of the larger early-2000 pages have nil problems jumping to the right place imediately
#
Zegnat s/load and jump/load then jump/
#
[tantek] right
#
[kevinmarks] http://epeus.blogspot.com/2002/11/#85630244
#
[kevinmarks] heh, these are so old they have <a name> rather than id
iasai joined the channel
#
[kevinmarks] webmention.herokuapp.com won't let me add a new domain, so I may need to make my own comment embedder…
#
Zegnat Maybe see if voxpelli us planning to open new slots.
#
sknebel [jdpinto1]: node-format-microformats is on NPM (the javascript package manager) and loaded as a dependency: https://github.com/voxpelli/webpage-micropub-to-github/blob/master/package.json#L50 - if you make a fork of it on on github and edit that, you can reference that instead by putting the git url into package.json instead of the version number
eli_oat, iasai, jackjamieson, snarfed and [jdpinto1] joined the channel
#
[jdpinto1] [sknebel] Thank you for your help and patience! I’ll have to take a closer look at this when I get the chance. Definitely headed in the right direction, though.
iasai, benwerd, ramin, [pfefferle], jjuran, cweiske, [kevinmarks], [tantek] and kisik21 joined the channel
#
Zegnat Alright. I need someone smarter than me. What does it mean when PHP says fopen() in a+ mode places “the file pointer at the end of the file”? I thought they meant cursor when they wrote file pointer, but ftell() still returns 0 :/
#
cweiske I think those are two different things
#
Loqi cweiske: [schmarty] left you a message 2 days, 22 hours ago: Looks like chat search is returning no results for all queries
#
cweiske the cursor is what you influence
#
cweiske a+ means it does not have to care how big the file is and thus does not load this information at all
#
cweiske it's just "append to the file" without caring about and of the information in the file
#
Zegnat But a+ allows for reading (unlike just a) according to the PHP docs, so surely the cursor position is important there?
#
Zegnat So the “write cursor” is set to the end of the file, but the “read cursor” to the start?
#
cweiske you are right
#
cweiske but:
#
cweiske In this mode, fseek() only affects the reading position, writes are always appended.
#
cweiske (says the docs)
#
Zegnat Yep. I was just thrown off by the fact that the “file pointer” is at the end, but ftell() is still telling me the cursor is at 0.
[asuh] joined the channel
#
Zegnat Ugh. I read now that fseek() is actually undefined even for a+ ... that documentation is just confusing! https://bugs.php.net/bug.php?id=76995
#
Zegnat is never going to understand these file pointer functions
eli_oat joined the channel
#
Zegnat I also do not understand why fseek()ing to the length of the file (thus putting the cursor behind the file’s last character/byte) doesn’t make feof() return true (end-of-file check). Need to take one more step.
iasai and eli_oat joined the channel
#
gRegorLove aaronpk, could you bump php-mf2 version in https://github.com/indieweb/rel-me/blob/master/composer.json or would that break things?
#
aaronpk I think that'd be fine
#
gRegorLove Cool. I'm updating the mf-cleaner lib in indiewebify.me and saw php-mf2 is at 0.2 because of that
krychu, chrisaldrich, iasai, [Khurt], eli_oat, [asuh], [cleverdevil], eli_oat1, [kevinmarks] and [jgmac1106] joined the channel