2016-10-03 UTC
kylewm and KevinMarks joined the channel
# 05:47 Zegnat KevinMarks: I do not want to replicate my hcard. Took the tantek route of using rel="author", and I think I used it right. But implementation seems iffy at this point. loicm joined the channel
# 06:10 KevinMarks You don't need to replicate it, you have all the bits there save a photo # 06:11 KevinMarks If you put p-author h-card on the link instead of u-author you'd get name and url # 06:21 Zegnat Maybe I'll do that when I get back from work. Maybe I will concentrate on permalink pages first KevinMarks_, loicm, KartikPrabhu, singpolyma, kylewm and tantek joined the channel
gRegorLove joined the channel
tantek joined the channel
# 18:55 aaronpk alright. super in progress, but who wants to try out micropub.rocks # 18:55 aaronpk i have a set of tests for creating posts in form-encoded and JSON formats # 18:58 barryf aaronpk I'll give the form-encoded tests a try. My JSON endpoint isn't far away so this will be very useful. # 18:59 aaronpk definitely let me know if you think any additional tests would be helpful # 19:11 barryf Nice. 100 and 101 pass, but 102 failed. I'll go and fix my numeric array offsets :) # 19:12 barryf Also, I first failed to read the fields correctly and added my URL rather than Micropub endpoint. May be good to have a way of editing/deleting endpoints. AngeloGladding joined the channel
# 19:37 bear hmm, I getting an "error=invalid_state" error trying to indieauth to my access endpoint to generate a token tantek joined the channel
# 19:38 bear i'm using my token generator to try and get a post token for micropub.rocks # 19:39 bear now that persona is gone, it's my only way to auth # 19:40 tantek bear, wait, pretty sure the servers still work til November # 19:40 aaronpk well you can always have indieauth.com email you a code # 19:40 bear but they are no longer presented as an option for indieauth # 19:41 aaronpk i figured there's no point in encouraging people to sign up for persona now if it's just going to go away soon # 19:41 aaronpk i still get a lot of new people using indieauth.com as an openid provider because it's one of the few left # 19:41 bear logging in again via clef is causing my indieauth code to cough up an invalid token error KevinMarks joined the channel
# 19:43 bear indieauth.com/auth is returning a 400 now :/ # 19:46 bear oh man, micropub.rocks is going to be *painful* ... 2nd test in and i'm already getting exceptions # 19:49 bear most of it is stuff I expect from specs that are written with a php-centric view of array encodings ;) # 19:49 bear ugh - and my json handling is blowing up # 19:50 aaronpk haha yeah, form-encoding is less a standard and more a PHP convention # 19:50 bear very much so - I was cringing when I saw micropub use so much of it # 19:50 aaronpk there unfortunately isn't anything very much better # 19:51 bear which is why I just sat on my hands and knew that the python versions would be very "flexible" KevinMarks joined the channel
# 19:54 bear hmm, my h-entry with two categories created the item but still failed # 19:54 bear that means the mf2 for the categories is wrong I guess # 19:55 bear oh - it doesn't scan the resulting code because it's a 202 I guess # 19:55 aaronpk it doesn't look at the post at all actually. the last box is always self-reported # 19:58 bear I would change the checkbox to read something like "Now check that the Post has both categories" # 19:58 bear the current wording implied that it would be verified by something, just now clear on who # 20:03 bear hmm, test 102 doesn't change the slug/title enough to cause a new post so it "passes" only because my code detects it's the same url and doesn't do a new create but rather an update # 20:04 bear right, and without a slug it's using the title as one KartikPrabhu joined the channel
# 20:04 tantek changing the slug slightly is a valid update use-case # 20:05 bear yea, this is the old discussion of why slug based metadata stores are fraught with problems - I get that # 20:05 aaronpk bear: that seems like it could lead to unintentional data loss # 20:05 bear since i'm the only one creating new items... :) # 20:06 bear my backend actually uses date+slug to minimize that # 20:07 bear i'm wondering if you should compare the location returned and fail it if it's already found in the domain's list of test passed # 20:08 bear so we can root out this slug == item once and for all # 20:08 bear which is going to cause me a *LOT* of reworking... # 20:14 bear laughs - photo-by-url and my code is telling me to basically to f-off ;) -- no signs of anything like a photo url in the debug output # 20:15 aaronpk i may need to double check i'm encoding everything right # 20:15 aaronpk i'm doing a somewhat crazy thing which is the text on the test page is actually what is sent to your endpoint. you browser extracts it and tells the server to go post it to your endpoint. # 20:21 bear aaronpk - that's better, my code still fails, but now I can see the entry that failed # 20:21 aaronpk cool. that was my fault for not URLencoding the photo URL # 20:21 aaronpk apparently php is more forgiving of things like that # 20:22 bear my list of items to look for comes from an old copy of the micropub spec, so I need to get it refreshed or realize that now any item can appear and make it more flexible # 20:22 bear that's why it's important to test with multiple implementations in different languages # 20:22 aaronpk i went with the approach of "store everything" and sort it out when displaying it # 20:23 bear yea, I filter everything and pass only known good on to the backend # 20:23 bear so if it's not in my whitelist the post creation code never sees it # 20:24 bear i'm going to sigh that deep sigh of php-inspired spec madness and carry on # 20:25 bear true, this one should have been "web dev inspired spec madness" # 20:25 bear but that's the world I live in now - i'm adjusting (and consuming more beer) # 20:27 bear section 3.3.5 is really targeting proxies or other intermediate agents - they should always store everything even if only acting on a small fraction of it # 20:27 bear because they don't know what the up/down stream handlers will want/need # 20:32 aaronpk takes a lunch break, considers having a beer and then getting back to work on the rest of the server tests AngeloGladding joined the channel
# 21:37 Loqi Ok, I'll tell them that when I see them next # 21:37 tantek Also curious about what other HWC co-organizers think of a checklist like that # 22:14 aaronpk okay finished the first "update" test. the site first creates a post, then you click "continue" to run the update request # 22:27 tantek I vaguely remember keeping latest 6 so that the IndieWebSummit for a given year would show up in the list of past events for that year during that year # 22:29 tantek right, that's probably good enough at this point
KevinMarks You don't need to replicate it, you have all the bits there save a photo
tantek.com edited /events/2016-10-05-homebrew-website-club (+51) "fix a little markup, ambiguate SF location pending co-host discussion" (view diff)
@bitworking Working on this myself, added HTTPS support, now slowly working on webmention support. https://twitter.com/waxpancake/status/783030608444297216 (twitter.com/_/status/783039385721638913)
aaronpk wonders what bear is going to say about this then https://micropub.net/draft/#unrecognized-properties
gregorlove.com edited /Instagram (-9) "/* Tweet text to Instagram */ updated instagram image URL, used photosrcalt template" (view diff)
tantek !tell chrisaldrich1 would a checklist like this help for IWC LA? https://github.com/donut-js/donut-js.github.io/issues/21
gRegorLove tantek: Think we should leave latest 3 months on /Events#Recent instead of 6? Or maybe just move them directly to /events/2016 and link to that from the #Recent subheading?