2015-07-19 UTC
j12t and gRegorLove joined the channel
# 00:07 kylewm there's basically nothing to it... just a manifest.js file that tells it which page to use
# 00:08 Jeena It would be cool if all could use the same manifest file and you'd set the domain on your own
snarfed and KartikPrabhu joined the channel
snarfed joined the channel
# 00:24 bear if I activate that (quill-firefox) where should it show up?
# 00:27 kylewm little airplane icon in the top right toolbar
# 00:28 kylewm or it might not be shown by default, in which case you have to go into "Customize"
# 00:29 aaronpk looks like it worked for me! minus a little fighting with my noscript plugin
# 00:29 bear oh I see - it uses the quill site but I could change it to mine if I had one
# 00:31 bear trying to use persona and persona's popup isn't fully loading
# 00:31 aaronpk huh, that's actually the one I used to sign in just now
# 00:35 bear the persona pop up loads but it has no fields and only a footer
# 00:38 aaronpk also you can sign in to quill outside the share thing
snarfed and frzn joined the channel
awolf joined the channel
# 01:05 Jeena bear I had the same problem just not with persona on this
KartikPrabhu joined the channel
snarfed joined the channel
# 01:27 GWG Anyone have any exciting Indieweb plans for the evening?
snarfed joined the channel
# 01:38 GWG ben_thatmustbeme: Yes, I think I caught it
# 01:40 aaronpk I think Loqi is catching the twitter expansion before the wiki hook
rbrt joined the channel
mdik joined the channel
# 02:04 GWG I have plans to write some post types code, but still percolating on design.
# 02:06 snarfed the irc log during the bridgy outage today is kinda hilarious :P
# 02:09 GWG snarfed: How do I ask you about the micropub stuff without pestering you?
# 02:12 GWG snarfed: I don't want to rush you. Just I paused the work so I would't have too big a dump on you.
# 02:13 GWG Next is Implied Post Types, Post Formats, and Post Kinds.
# 02:13 snarfed GWG: no need to pause work! you can make a new branch from the PR and keep working
# 02:15 GWG snarfed: I know. I just figured I'd give you time. Again, trying to avoid pestering.
scor joined the channel
# 02:18 GWG I figured I'd work on the Post Kinds side, with code I'll use later to interface with the Micropub plugin
schmarty, rbrt and j12t joined the channel
# 02:25 snarfed i haven't kept up with micropub changes since i wrote this
scor joined the channel
# 02:26 GWG snarfed: I found a few things that people were doing that were non-standard as well.
# 02:30 snarfed thanks Jeena! i'm actually mostly interested in the non-auth parts
# 02:31 aaronpk snarfed: yes it is! and actually reflects my latest implementation of a micropub endpoint as well
# 02:31 aaronpk the only part that is somewhat in flux is the "mp-syndicate-to" bit
# 02:32 GWG snarfed: I implemented syndicate-to and made mp-syndicate-to an alias to it. Easy to swap the other way
cmhobbs joined the channel
# 02:41 GWG snarfed: I forgot to merge the commits when I made the change.
# 02:41 snarfed …oh you did it in master. that'll still work, but you'll want to use a branch next time
# 02:41 GWG snarfed: I did. I switched the whitelist to showing everything.
# 02:42 GWG snarfed: I know. I don't do much branching. I have to learn to work better with others.
j12t joined the channel
# 02:47 GWG snarfed: I just realized storing all variables means it stores the access token.
# 02:48 snarfed also GWG if you plan to keep sending PRs, i'd love to see a unit test suite eventually!
# 02:48 GWG snarfed: I'll have to figure out how to write one of those.
schmarty joined the channel
# 02:49 GWG My approach so far has been testing using micropub clients
# 02:49 GWG That's why I filed an issue against Quill
# 02:51 GWG That's how I realized that having a query string for your endpoint means that querying it is an issue.
# 02:51 GWG ben_thatmustbeme was able to fix that in Postly already
KartikPrabhu and j12t joined the channel
# 03:28 GWG ben_thatmustbeme: Just was more annoyed it took me that long to figure out what the issue was so I could file it.
j12t joined the channel
# 03:42 snarfed GWG: curious, how did you decide to return the q=syndicate-to response as url-encoded with php array syntax?
# 03:42 GWG I thought that was what it was supposed to be?
# 03:44 GWG It seemed to work in testing with Postly, as the only client that works with queries.
# 03:44 snarfed oh it may be! i don't know. i'm asking how you know :P
# 03:45 GWG Also, everything else seemed to be urlencoded, so I figured..
# 03:45 snarfed i don't actually see much at all specified about responses, right? definitely not url-encoded?
# 03:45 snarfed could you and aaronpk and ben_thatmustbeme agree on that and add it to the spec?
# 03:46 GWG ben_thatmustbeme: What is your take?
# 03:48 ben_thatmustbeme ahh, yes, i don't think i'm the only one to do that. I think aaronpk does it too, many require a token to get that though
# 03:49 GWG aaronpk returns everything as content type application/x-www-form-urlencoded
# 03:50 snarfed (and the expected param name, and that it's php array syntax, etc)
# 03:50 GWG I think the array syntax is a definite. Everything else is
# 03:53 GWG There was an example in the brainstorming section
schmarty and schmarty1 joined the channel
# 04:12 snarfed thanks for working on this btw, GWG! i definitely appreciate the updates
# 04:12 GWG snarfed: I'm incorporating your feedback.
# 04:15 GWG It won't store the access token anymore
nedsauce joined the channel
# 04:19 snarfed GWG: looking good! i think one last round of feedback, and one last round of manual testing to confirm everything still works after the changes, and it'll be good to go
scor joined the channel
# 04:28 snarfed GWG: could you also document the new filters and mention them in the changelog?
# 04:29 GWG snarfed: Anything else that needs to be blacklisted by default?
# 04:29 GWG I can't think of anything right now.
# 04:31 GWG I have one more thing I probably should add. Pushing now.
# 04:49 snarfed GWG: thanks! all i have left is a few small readme tweaks
# 04:49 GWG snarfed: I'll be back next week with 0.5
# 04:50 snarfed mind manually testing everything again, both new and existing features? just to make sure we didn't break anything with the review feedback
# 05:00 snarfed great! i'm guessing editing with u= doesn't work just yet though
# 05:03 GWG I think it only shows url= right now
# 05:07 GWG snarfed: I think I reverted the u= change because the spec is unclear on it
# 05:10 GWG That is why I got confused about it
# 05:11 snarfed seems like it just changed from url= to edit-of= . not too confusing
# 05:16 snarfed GWG: thanks! we still want to be backward compatible w/url= though
# 05:21 GWG Fixed the typo, added the conditional
# 05:21 GWG i will have to test editing a bit more.
# 05:23 snarfed looking forward to the unit tests before the next PR! :P
# 05:23 GWG snarfed: As soon as I learn. Is this PR done though?
# 05:24 snarfed have you manually tested url=, edit-of=, and q= ?
# 05:24 GWG I haven't done any editing at all
# 05:27 GWG edit-of does not work as per spec. It may never have
# 05:29 snarfed can you elaborate? do you mean the update, add, and delete params don't work?
# 05:29 snarfed if so, that's ok, we just want edit-of to work like url currently does
# 05:30 GWG But the url one didn't work properly
# 05:30 GWG It overwrites the published time.
# 05:31 snarfed ah. that was an existing bug? that's fine, we can fix that later
# 05:31 snarfed otherwise updating with url= and edit-of= both work?
# 05:31 GWG Yes. It also overwrites previously set fields if the new ones are blank.
# 05:32 snarfed huh, i thought the PR prevents writing blank values
# 05:34 GWG I'd suggest reading the existing post out, and merging the two arrays.
# 05:35 GWG That is for post meta, not for the post args
lukebrooker joined the channel
scor joined the channel
# 06:08 GWG Next, I have to figure out a simple way to write out the massive list for implied post types.
lukebrooker joined the channel
# 06:09 snarfed GWG: maybe unit tests before that? they'll let you skip all the manual testing :P
# 06:10 GWG I have to learn how to write those. Which means I have to find some literature.
# 06:10 GWG One is a problem I can think about, the other is a problem I can't think about till I have some more knowledge.
# 06:12 snarfed true! but the tests will give us both more confidence that everything keeps working through the rest of your upcoming PRs
# 06:12 snarfed especially since i don't actually use micropub much, so i wouldn't notice otherwise if it breaks
# 06:13 GWG snarfed: My intention is to get it to the point I could use it.
# 06:14 GWG On that note, I'm going to sleep on i
loic_m and evalica joined the channel
awolf, evalica, eschnou, catsup, tvn, loic_m, KartikPrabhu, schmarty, scor and frzn joined the channel
catsup joined the channel
catsup, loic_m, lukebrooker, evalica, tvn, eschnou and KartikPrabhu joined the channel
catsup and evalica joined the channel
# 13:45 GWG ben_thatmustbeme: Why do you send title as opposed to name, which is in the examples?
catsup joined the channel
catsup joined the channel
# 13:52 seekr waves at ben_thatmustbeme and hopes to chat about a f2f meeting soonish (?)
evalica joined the channel
# 13:53 GWG ben_thatmustbeme: Want me to file an issue?
# 13:54 GWG ben_thatmustbeme: What does your client use for editing? Edit-of?
evalica, catsup and snarfed joined the channel
# 14:17 GWG ben_thatmustbeme: The spec no longer shows url=, it shows edit-of as the reserved property.
fkooman, catsup and evalica joined the channel
# 14:23 ben_thatmustbeme is getting annoyed with things being added to the "spec" without the reasoning behind them
evalica and catsup joined the channel
evalica, Pierre-O and catsup joined the channel
# 14:33 voxpelli I guess the Micropub spec is still pretty young so things should be expected to change as more experience is gathered from additional implementations?
catsup joined the channel
# 14:36 voxpelli I rarely have time to read any of the logs, I just keep working in my part of the indieverse and sync up with the rest every now and then :)
# 14:36 GWG I need to figure this out, I guess
# 14:37 voxpelli but a good argument to get automated test tools, so we can all easily check whether we need to change something in our endpoints or clients
evalica joined the channel
# 14:38 GWG voxpelli: I'm not sure I'm the best person to write them.
# 14:38 voxpelli GWG: I may take a stab at it, but not sure I have the time to get something out soon
# 14:39 GWG I just keep discovering different issues.
# 14:39 GWG For example, now this editing issue.
# 14:39 GWG Who, other than ben_thatmustbeme, supports editing in their client?
evalica and catsup joined the channel
# 14:41 GWG ben_thatmustbeme: My problem keeps being that I think I'm making a mistake, when I'm coming up against differences in implementation
# 14:41 GWG Which aren't anyone's fault, but confusing me.
# 14:42 voxpelli ben_thatmustbeme: your entry on the client list on the micropub list doesn't say that you support edits btw
# 14:43 voxpelli I find the currently documented edit API a bit odd – because the form-encoded values all of the sudden mimics the JSON-format rather than the format of form-encoded creates
# 14:44 GWG snarfed: My testing is stymied by this issue though.
# 14:44 GWG I think we need to agree on the edit API before I try to improve edit support.
# 14:44 GWG So, maybe I should just focus on create right now
# 14:44 snarfed GWG: i haven't followed what "this issue" is, but adding unit tests to the existing code should be doable without solving any new issue
# 14:44 voxpelli GWG: by the edit issue? if so: hard to test something that it seems that no one really supports. If there's no support for a spec, does that spec really exist? And can one test something that doesn't really exist? ;)
# 14:45 snarfed heh, we're just talking about unit tests for existing code. (or at least i am.) voxpelli you're probably talking about something bigger that you can run against any server or client
# 14:46 voxpelli snarfed: yep, as that's the only way we can ensure conformance between many different implementations
# 14:46 GWG snarfed: The issue is still the same. ben_thatmustbeme is the only one supporting edits I've noticed so far in his client, and his implementation doesn't match the one in the wiki.
# 14:46 GWG So, trying to test we're accepting the right requests is hard for that reason.
# 14:46 snarfed GWG: sure, understood. worth solving! independently, it'd be worth writing unit tests. they're unrelated.
catsup joined the channel
# 14:47 snarfed even if we just added unit tests for all the non-edit functionality, that'd be hugely valuable
# 14:48 voxpelli my endpoint currently solves "editing" by simply replacing any existing document if someone tries to create it again
# 14:49 GWG ben_thatmustbeme: That is what your edit functionality does as well
# 14:50 voxpelli ben_thatmustbeme: I just create a new file from scratch and let that file replace the old one – no pulling of data – so the client has to have persisted some stuff in itself, like the Editorial.app workflow I created does
tilgovi and catsup joined the channel
# 14:52 ben_thatmustbeme voxpelli: i don't pull from the db directly, what i do is passing a url to the client, will have the client poll and parse the mf2 of the post and then prefill the fields
# 14:53 voxpelli ben_thatmustbeme: right, that makes sense on the client side, and then just like I'm doing – make a replace on the server side
catsup joined the channel
# 14:58 voxpelli really? dropping all mp- prefixes? I thought such prefixes made sense, to differentiate between directives made to the micropub endpoint itself and directives that are part of the actual content
# 14:59 aaronpk the edit syntax is derived from the idea of "edit posts"
# 14:59 voxpelli In my endpoint I include all non-mp prefixed data in the post files
# 14:59 aaronpk i thought i had an example of showing that derivation on the wiki, but can't find it
# 15:02 snarfed voxpelli: ooh perfect thank you! does it set up test db tables too? or mock the db? or…?
catsup joined the channel
# 15:03 aaronpk i went back and forth on this with tantek a bunch
# 15:03 voxpelli snarfed: not sure – I know the guy that has created those repos, but never actually used them myself – I think he will happily answer questions (but he may be on vacation now of course)
# 15:04 voxpelli aaronpk: on the question that GWG had: does any client actually implement these edits? so one can have a client to try with?
# 15:05 aaronpk voxpelli: i don't think so. i was testing my development of a micropub endpoint with curl commands
# 15:06 voxpelli ben_thatmustbeme: edits makes use of the mf2 JSON-data format
# 15:06 aaronpk the idea is you create an "edit" post which describes what changes in the thing you're editing
# 15:07 aaronpk you don't actually have to create the edit post. the micropub syntax for edits is derived from this "edit post", basically a trimmed down version of it, similar to how creating a post is a trimmed down version of an mf2 h-entry
# 15:08 ben_thatmustbeme "Consumers of this post must act only on non-empty property values on the object embedded on the "p-update" property. " made me think you are proposing this h-entry to cause the edit
# 15:09 aaronpk and the nice thing is that you can then use the same logic to handle parsing an edit post and parsing a micropub request
# 15:09 voxpelli I think I like this (would still keep mp- prefixes for eg. syndicate-to and such though I think)
# 15:09 raucao was there any progress with private sharing (groups or sth else) after Düsseldorf?
# 15:09 aaronpk so you are one step closer to being able to accept "pull requests" from edit posts
# 15:09 voxpelli aaronpk: except that edits use the JSON-data format in a form-encoded format which makes it pretty different from a normal form-encoded micropub-reqiest, no?
# 15:10 aaronpk voxpelli: yeah the challengs is there isn't a good non-nested form-encoded syntax for edits
# 15:10 voxpelli aaronpk: for consistency I would change the normal creates to the same format as well
# 15:11 voxpelli right now I represent everything as the JSON objects internally in my endpoint, and convert all form-encoded object to the JSON-style
# 15:12 voxpelli now I would have to expect two different formats for the form-encoded formats
# 15:13 aaronpk form-encoded create requests have all properties of the post in the request body, except the "h" value. json-encoded create requests have all properties under the "properties" key
# 15:13 aaronpk my endpoint actually has the same code handling the create logic, i just pass it either $_POST or $request->properties
# 15:13 voxpelli yes, and form-encoded edits also have all properties under a properties key then?
# 15:14 aaronpk form-encoded and json-encoded edits use the same serialization
# 15:15 voxpelli ben_thatmustbeme: all my micropub actions are added as commits to a git branch somewhere ;)
# 15:15 snarfed i think wordpress may also let you propose line-level changes to posts. not sure
# 15:16 ben_thatmustbeme i'm just thinking if facebook allowed people to 'suggest an edit' how insanely terrible it would be
# 15:16 voxpelli ben_thatmustbeme: could actually be used by an editor to pull content live from authors
# 15:16 snarfed ben_thatmustbeme: true! fortunately we can learn from more silos than just FB :P
# 15:17 ben_thatmustbeme snarfed: s/facebook/twitter s/facebook/google+ s/facebook/anything that isnt' editing code
# 15:17 snarfed code + docs + posts is a pretty damn big space :P
# 15:18 snarfed hoo boy, migrations. always fun/painful. break a leg!
# 15:19 ben_thatmustbeme he put everything in a seperate DB that had nothing to do with the original, and he clearly did not know how to design a DB.... or organize code
# 15:24 ben_thatmustbeme snarfed: also not any sort of straight migration.. a whole bunch of ruby code to translate structure
# 15:24 ben_thatmustbeme and consume things like changes to customer data being stored a html in the db...... rather than fields and generating the html on output
scor joined the channel
# 15:53 bret So in micropub, where you accept a JSON body instead of urlencoded
# 15:53 bret there is no suppored way to handle files?
# 15:53 bret it dosnt really make sence to shove JSON into multipart
# 15:54 bret eg, nobody is supporting base64 encoded images or anything like that?
# 16:02 bret there might be better multipart handlers
awolf and chreekat joined the channel
# 16:17 KevinMarks_ data URLs inside the content body work. kinda sketchy, but they do work.
# 16:18 bret KevinMarks_: you mean like a URL pointing to an image hosted elsewhere?
# 16:18 KevinMarks_ you could check for them and put them in your own storage instead...
# 16:18 KevinMarks_ no I mean an inline data: URL for an image. Quill editor sends them
tantek joined the channel
# 16:19 bret KevinMarks_: oh hrmm i haven't inspected the Quill Article editor
# 16:20 bret KevinMarks_: are they based64 encoded?
# 16:23 voxpelli well, data:uri:s in image tags is more a kind if HTML-data than file data :P so I will probbaly leave them alone
schmarty joined the channel
# 16:23 voxpelli multipart is better as one can stream upload those directly to disk or cloud service without ever keeping them in memory or such
# 16:24 bret is there an S3 or flicker Writable Stream package?
# 16:24 voxpelli what I did at my previous Busboy-project was that I streamed the upload directly to Cloudinary
KevinMarks joined the channel
# 16:26 bret voxpelli: did file.pipe(stream) not work?
# 16:26 voxpelli bret: not sure, wrote that 2 years ago + haven't done that much with node's streams
# 16:27 bret i think that might be some suguar to make streams more streamy
# 16:28 bret KevinMarks_: the alternative to is just embedding external images on authorship
ttepasse, snarfed, Rev_Illo and mdik_ joined the channel
# 16:54 GWG returns from lunch and catches up
Rev_Illo joined the channel
raucao joined the channel
schmarty and awolf joined the channel
# 17:25 GWG I keep sending snarfed pull requests over unit tests. I'll get there, but I just fixed some nagging behaviors.
# 17:28 GWG ben_thatmustbeme: Can I send you a snark about my feelings on snarks?
# 17:29 kylewm thinking about person tagging... i'm considering "upgrading" any regular tag to a person tag if the name matches the name of a contact
# 17:29 aaronpk Yeah I definitely don't want to go down the route of trying to do file uploads with JSON requests
# 17:30 kylewm any raeson that shouldn't work? like "tantek" always gets upgraded to an h-card
# 17:30 tantek kylewm: that sounds like a reasonable UI - but I wouldn't do it serverside
# 17:31 tantek such person-name auto-suggest UIs are fairly common on various silos, yet they always give you the ability to just have the plain text
# 17:31 kylewm so server side you'd recommend storing it more explicitly
# 17:32 kylewm rather than the implicit "if these two things happen to match up"
# 17:32 tantek right - I don't think the server should try to be "intelligent" like that, but rather build such intelligence into the interactive client so the user can make an explicit choice (and have it be respected by the server)
# 17:32 aaronpk I was struggling earlier trying to decide whether to store person tags in a separate list from regular tags
# 17:33 tantek no need - as you can store them just as URLs alongside non-URL tags
# 17:33 tantek if you like, you can *cache* the person-tagness (name, photo etc.) next to the URLs
# 17:33 aaronpk I can see doing it both ways. I decided to store them separately so my logic in the templates is simpler
scor joined the channel
# 17:35 kylewm do either of you cross-reference tags to your contacts-list thing?
# 17:39 GWG Post Kinds thinks everything coming in from Micropub is an RSVP. Think I need to switch projects for a bit.
# 17:39 aaronpk kylewm: I don't yet but only because I was in a rush doing it and only show the URL right now. Planning on looking up the URL in my nicknames cache to display the tag better soon
# 17:41 emmak what about uploading photos and media separately from the post?
# 17:42 kylewm has decision paralysis, will come back to this later
frzn joined the channel
# 17:45 aaronpk emmak: there should be no problem referencing a URL in the "photo" property
# 17:46 aaronpk If you don't do anything special, and just store Micropub values as they come in, it will Just Work
schmarty joined the channel
# 17:51 GWG aaronpk: You still need to know how to display the properties, don't you?
# 17:51 aaronpk Yes but that can happen when the post is being rendered
# 17:51 emmak a client could publish a jpg binary, save the returned "location" header, then make a json post with photo property set to that location
# 17:53 kylewm presumably that's the kind of think you'd want to do to support storing photos in S3 or something
# 18:00 aaronpk Yep I expect some clients/servers to do that for sure
# 18:00 voxpelli emmak: I think I saw somewhere around Micropub that one should support plain URL:s as input in addition to file uploads
# 18:01 voxpelli then how one makes a file available on that URL is up to oneself more or less
# 18:01 aaronpk I was actually considering making the Quill editor do that, rather than the crazy data uri thing it does now
# 18:01 voxpelli I'm creating supports for files mostly because OwnYourGram is sending them and it felt like a good tool to support :P
# 18:05 aaronpk data uri is a syntax for embedding file data (usually images) in an HTML img tag
# 18:08 emmak that is cool you can start a wiki entry just by talking to loqi
# 18:13 KevinMarks_ hm, if a micropub endpoint parses for img URLs, fetches them and stores them in its own store, then using data: or an intermediate upload service ends up equivalent
squeakytoy2 joined the channel
# 18:25 tantek unintended typo: interopeerable - interoperable peer to peer
# 18:32 bear that is when I met most of you for the first time
# 18:34 tantek I think I'm going to leave that typo in just for the subtle lol
# 18:42 GWG I remember the day I met bear and tantek.
# 18:55 tantek !tell kylewm been playing with Woodwind, noticed a few things. 1) subscribe to a domain (kylewm.com) errors w/o http:// in front, 2) long URLs in posts causes horizontal scroll, 3) could use white-space:prewrap for displaying notes with higher fidelity.
# 18:55 Loqi Ok, I'll tell them that when I see them next
# 18:59 Loqi Ok, I'll tell them that when I see them next
# 19:01 kylewm tantek: cool thanks for the feedback! definitely want to avoid horizontal scroll. I saw those errors in the logs trying to subscribe to schemaless domains. I thought <input type="url"> was doing input validation on those for me, does noscript bypass that?
# 19:01 Loqi kylewm: tantek left you a message 6 minutes ago: been playing with Woodwind, noticed a few things. 1) subscribe to a domain (kylewm.com) errors w/o http:// in front, 2) long URLs in posts causes horizontal scroll, 3) could use white-space:prewrap for displaying notes with higher fidelity. http://indiewebcamp.com/irc/2015-07-19/line/1437332112069
# 19:04 tantek not all browsers support input type="url", and you must always do input validation on the server regardless since form submissions can be faked
# 19:04 kylewm ok good to know. what browser were you using out of curiosity?
# 19:05 tantek I'm not a huge fan of the draconian UI imposed by input type=url FWIW
eschnou joined the channel
loic_m, KartikPrabhu and scor joined the channel
# 19:38 KevinMarks I should blog the fragment links to things with floating headers thing
MMN-o, KartikPrabhu, schmarty, squeakytoy3 and mapkyca joined the channel
schmarty1, MMN-o, cweiske, chreekat, eschnou and MMN-o_ joined the channel
# 20:31 KevinMarks_ if I have a website that is static pages, should i mark it up as h-entry?
KevinMarks__ joined the channel
# 20:36 kylewm grrr, i have some python code for calling the flickr api. it works fine on the command line, but when i run it inside appengine it gives Permission denied
KartikPrabhu, MMN-o, snarfed and g3funk joined the channel
# 20:51 cweiske shouldn't that be yourpythonflickrapi-- or flickr-- instead of bashing a protocol?
# 20:54 Loqi slack/snarfed: it seems pretty generally accepted that oauth1 is harder to use than oauth2
MMN-o joined the channel
# 20:55 Loqi slack/snarfed: not "bashing" necessarily; oauth1 was a huge achievement and step forward!
# 20:55 GWG snarfed, saw your message, by the way
ttepasse_ joined the channel
# 20:55 kylewm cweiske: maybe, the fault could be with python or flickr or whatever. but oauth1 is what makes it opaque and difficult to debug
MMN-o and chreekat joined the channel
interactivist, MMN-o, KartikPrabhu, scor and ttepasse joined the channel
# 22:00 GWG pfefferle just added me to wordpress-uf2. I wonder if this means I have arrived.
KartikPrabhu and snarfed joined the channel
# 22:08 KevinMarks__ Appengine fetching things needs care, I was getting some errors with unmung.com until I switched to explicit fetching
snarfed1 and lukebrooker joined the channel
# 22:27 GWG Learning to write unit tests is hard
# 22:27 bear it's hard but oh my is it so useful and helpful
# 22:28 snarfed every time you write code, you then have to test it. you can do it manually or automatically. i know which i prefer :P
# 22:28 snarfed GWG: are you looking at the examples in the issue i filed?
# 22:28 bear start simple and borrow heavily from other people's use of the unit test lib you use
# 22:30 snarfed GWG: other than phpunit, the two key parts for these will probably be 1) either creating/deleting the db tables for the test to use, or mocking out the db, and 2) figuring out how to give the wordpress code an http request and get a response
# 22:30 snarfed i definitely recommend looking at the example links in that issue
# 22:30 bear you will find working with databases and HTTP endpoints some of the hardest tests to write
# 22:31 snarfed maybe so. fortunately in this case we can just copy what these examples do
# 22:31 bear I will often isolate the code that has to write to the database just so I can test everything around it except the actual database call
# 22:31 bear ah - having working unit tests to copy from is awesome sauce
tvn joined the channel
KartikPrabhu and tantek joined the channel
# 22:40 GWG snarfed: The thing is that it is trying to test outgoing code, not incoming code.
# 22:40 snarfed and which of those three linked repos are you talking about?
# 22:41 GWG snarfed: I think I have to write something that simulates a client.
# 22:41 GWG In the unit tests I saw there, they are unit testing the client code. Whereas we're unit testing server code
# 22:42 snarfed huh, ok. let me look. i'm sure at least one of those three examples will work for us
# 22:43 GWG I think I need to write a class to generate Micropub calls, and then run unit tests with that so it can test the server.
# 22:51 GWG Either way, I'll try and learn, I suppose. Having a class that implements client functions is not a bad thing either.
# 22:53 snarfed anyway, don't go too far down that road. we're not looking for a test client; we just want unit tests
# 22:53 GWG snarfed: None of them note edit support.
# 22:53 GWG snarfed: That was the issue earlier today
# 22:54 GWG snarfed: So basically, my idea is to write some functions that implement a barebones client, then the unit tests use them to send in some arbitrary data.
# 22:54 snarfed GWG: i understand the idea, i just don't think that's what we want. give me a little time and i'll find the right example.
# 22:55 GWG snarfed: Okay. As I said, might be the wrong approach.
tantek and friedcell joined the channel
# 22:55 snarfed to be fair, GWG, that is basically what we're going to do, i just expect it will be much more minimal than you're thinking
# 22:57 GWG snarfed: I figure...if I have to write the code, might as well write it so it can be reused, even if barebones.
# 22:58 snarfed GWG: sure! if you want to write a general purpose micropub tester that can be used against any (remote) server, sure. that's not what i'm asking for here though
# 22:59 bret aaronpk, kylewm (maybe even snarfed) whats a good example of a JSON web token signing secret?
# 23:04 GWG snarfed: I know it isn't strictly necessary, but I was thinking of the big picture.
# 23:05 snarfed GWG: not at all, it's a great idea! i know voxpelli was interested earlier. i'm sure lots of other people would be too
# 23:05 snarfed i just want simple unit tests for wp-micropub too, independent of a bigger test client
KevinMarks joined the channel
# 23:07 GWG snarfed: What is missing, from what I can see, is a micropub client library
# 23:07 GWG snarfed: Did you find the simple example though?
# 23:08 emmak bret: long enough that its impractical to brute-force
# 23:08 snarfed GWG: not yet, working on day job stuff. i'll let you know.
ehlovader joined the channel
# 23:10 emmak 8 bytes would be 2^64 possible secret keys
KartikPrabhu joined the channel
# 23:11 emmak i've used 16-18 bytes from the openssl random bytes function for this type of thing
# 23:15 GWG snarfed: p3k-micropub is part of aaronpk's p3k system, and thus pretty specific. Also appears to be server, not client.
# 23:16 snarfed GWG: sure, ok. the taproot one says it's client side
friedcell joined the channel
# 23:16 GWG snarfed: It appears to be authentication only.
# 23:16 snarfed GWG: so "implements both client app and resource provider authentication " on the wiki is lying?
chreekat joined the channel
# 23:18 aaronpk My rule of thumb with keys is as long as you have at least as many bits as a uuid you're fine
# 23:21 GWG aaronpk: Do you know if Taproot ever got its full micropub client library? I can't find it
# 23:22 GWG Because the one linked on the page is an Indieauth client library, not a MicroPub one
awolf and friedcell joined the channel
# 23:25 GWG barnabywalters put it in, which is consistent
# 23:26 tantek is trying to follow the auth related discussions as best he can.
gRegorLove joined the channel
snarfed joined the channel
tantek and KartikPrabhu joined the channel
# 23:52 GWG I use him as a reference all the time
# 23:52 GWG Did you catch his talk at Loopconf on how to avoid breaking backward compatibility when changing your data structure?
# 23:57 GWG snarfed: Neither do I on conferences. But they put the videos up. I watched them while doing something else. At least the interesting ones
# 23:57 GWG After I contributed to breaking acegiak's site, it was a topic I had strong feelings about.