2013-11-15 UTC
_6a68 and tantek joined the channel
# 00:14 bret jekyll does future posts, but its a simple date sort, nothing more
# 00:14 bret you can prevent future posts from rendering too
_6a68 joined the channel
# 00:16 bret also, in flat file applications, are you litterally reading a post file everytime someone requests to look at it?
# 00:16 bret or do you process those files into ram or something?
# 00:19 aaronpk p3k processes the flat file and caches it in RAM. not the whole HTML page, just the HTML after building the page's content. (header/sidebar/footer stuff is not cached along with the page)
tantek joined the channel
# 00:29 tantek flakey network + log caching = am I speaking to myself or not
# 00:30 tantek so does that mean that jekyll renders future posts by default? instead of not showing them before their time by default?
# 00:31 tantek bret - in answer to your question, Falcon is literally reading a post file every time someone requests to look at it. Assuming "typical" OS file caching that keeps commonly read files in memory, this is not a problem. See http://indiewebcamp.com/Falcon#Storage_format for details
npdoty, scor and ryana joined the channel
# 00:46 bret tantek: sorry had to move locations for a bit
# 00:48 bret jekyl just renders static pages. It can render future posts, or it can ignore them. default is ignore
# 00:50 bret plus personal sites tend to be fairly low traffic, at least mine is
ozten, ryana, thatryana, _6a68 and josephboyle joined the channel
# 01:22 tantek hmm - now I have think about how to write good time code in CASSIS (date stuff is taken care of pretty well)
bnvk, caseorganic and tantek joined the channel
# 01:46 tantek and this is where I consider regretting not storing my datetimes in UTC
# 01:51 tantek well I store(d) my datetimes in the more easily verifiable *human* local time
# 01:51 tantek but basically stuck to Pacific Time since no matter where I travel I keep track of that personally
# 01:52 tantek so now I just need to write code to convert that (reliably) to UTC which will be - interesting (given DST etc.)
# 01:55 tantek it's an interesting design challenge. but I think the better option is to keep the data as easily human verifiable as possible.
# 02:07 tantek writing code for easier debugging, even if it means more code
# 02:07 tantek knowing that more time is spent debugging in general
tpinto, scor, squeakytoy, skinny, snarfed, tantek, b0bg0d, bnvk, squeakytoy2 and josephboyle joined the channel
# 05:56 KevinMarks_ !tell tantek thanks for the twitter length functions. Noterlive needs them
# 05:56 Loqi Ok, I'll tell him that when I see him next
brianloveswords, tpinto, npdoty, cweiske, vrypan, bnvk, b0bg0d, andreypopp, poppy, friedcell, icco, Jihaisse and pfefferle joined the channel
tantek joined the channel
# 08:18 Loqi tantek: KevinMarks_ left you a message 2 hours, 22 minutes ago: thanks for the twitter length functions. Noterlive needs them
LauraJ joined the channel
tpinto joined the channel
# 08:23 tantek pfefferle not that I know of - he just asked on Twitter
# 08:23 pfefferle tantek ok... will ask him to join the wiki, he can perhaps help me writing a page about theming
bnvk joined the channel
# 08:37 tantek and I just got posting to the future working :)
# 08:37 tantek I can write posts, post-date their published date, and they don't show up on my home page until their datetime passes.
LauraJ joined the channel
# 08:45 tommorris Jihaisse: have to wait a few days to find out. Postal system and all.
# 08:48 tantek just got the code flow to properly handle dt-published values that are in the future
# 08:48 tantek it's also good foundational code for "future" posts like events and travel plans
tpinto joined the channel
tantek, friedcell, bnvk and b0bg0d1 joined the channel
# 09:41 KevinMarks_ my liking for making things in node goes down with each deployment wait on heroku. Thinking that this isomophic javascript thing isn't worth the setup
bnvk, cweiske, skinny, melvster, barnabywalters, pfenwick, tpinto, friedcell, vrypan and andreypopp joined the channel
pfefferle_, bnvk, pfefferle, skinny, andreypopp, squeakytoy, LauraJ, scor and brianloveswords joined the channel
andreypopp and indiewebcamp-vis joined the channel
andreypopp_, bnvk and barnabywalters_ joined the channel
andreypopp, skinny, snarfed, bnvk, tantek and ozten joined the channel
skinny, scor, ryana, thatryana, snarfed, caseorganic, fmarier, squeakytoy, tantek, bnvk_, bnvk__, LauraJ, bnvk, andreypopp, paulcp, smus, KevinMarks2, warden, Slyphoria, KevinMarks_, XgF and barnabywalters joined the channel
# 20:02 barnabywalters aaronpk: do you strip the protocol off URLs when saving into your mention archive?
# 20:04 aaronpk I strip off the protocol when building the path on disk to save the parsed data, but I still store the full URL of mentions
# 20:04 barnabywalters okay — and when an update comes, do you overwrite the original (presumably the history is stored in git)?
# 20:04 tantek is this something to capture as part of indiearchive?
bnvk joined the channel
# 20:06 barnabywalters aaronpk: sounds good. For my use cases I want to be able to easily list and jump between different copies of archived URLs, so I’m making the path into the folder, and then storing EPDSSS.html under that
# 20:06 aaronpk don't double slashes in URLs and filesystem just get collapsed into a single slash automatically?
# 20:07 aaronpk must be a browser and unix tool thing then, not a filesystem thing
# 20:09 barnabywalters this way I can always easily get the latest version of a mention, but list previous versions
# 20:10 tantek I'm a little unclear on what scenario your having problems with.
# 20:11 barnabywalters tantek: referring to the double-slash thing, or the listing-previous-versions thing?
# 20:11 tantek this, what does this mean? (i.e. explain the use case that causes this problem) "urls with EPD/SSS/https://domain.com aren’t filesystem-friendly"
# 20:12 aaronpk I think he's talking about storing the arcives as files on disk with the same folder structure as the URL
# 20:12 barnabywalters I didn’t know that double slashes were collapsed into single ones, so that case could be dealt with by merely replacing https?:// with https?:/
# 20:13 aaronpk barnabywalters: I don't know what exactly is doing that, but if you type cd /var//www you'll get to the right place
# 20:13 tantek btw - TimBL has since said that had he known better, he would have omitted the "//"
# 20:14 tantek no it can't since "httpsaaronparecki.com" is a valid domain
# 20:14 tantek aaronpk - there's the general problem of URL-safe characters being maybe a superset of filename path-safe characters
# 20:15 XgF aaronpk: Per POSIX, double slash can be collapsed everywhere*except* at the start of a path. // and / may point at different locations
# 20:15 tantek so if you're putting a URL in your filesystem, you need to handle that general problem
# 20:15 tantek rather than case by case dealing with ":" etc.
# 20:15 XgF aaronpk, barnabywalters: https: is a valid path everywhere but Windows?
# 20:15 XgF aaronpk: Dunno. Because PHP is an epic pile of fail? :-)
josephboyle joined the channel
# 20:17 tantek XgF - hence why those of us that use it use a deliberate subset :)
# 20:17 XgF aaronpk: Hey, I use PHP occasionally too. It's practical, maybe pragmatic, and also disgusting :-)
# 20:19 XgF For the curious, POSIX allows // and / to differ because I think Apollo used // to indicate "other machines" - think similar to Windows UNC paths
# 20:19 barnabywalters “The double slash, though a programming convention at the time” — was // a “programming convention”?
# 20:20 XgF barnabywalters: Maybe taken from Apollo (see my above comment on POSIX)?
# 20:21 XgF Hmm, does anyone know if NeXTSTEP did something similar? TBL developed HTTP/HTML on a NeXT cube
# 20:23 tantek and then we could have had protocol relative URLs by starting them with ":" rather than "//"
# 20:27 barnabywalters aaronpk: do you do anything special to handle query string params when translating URL -> filesystem path?
# 20:31 tantek or maybe just drop webmention sources with ".." in the URL
tpinto joined the channel
# 20:32 tantek since we're talking about auto-ignoring spam anyway
# 20:32 tantek might as well auto-ignore such sketchiness as well
# 20:32 barnabywalters yeah, there’s probably no good reason to support mentions from people with .. in the URL
# 20:32 XgF Though still, you probably want an escape sequence for query strings and such
# 20:33 XgF Maybe "__", with "__" -> "___", "?" -> "__q" or similar?
# 20:33 tantek I'd say start with blanket dropping such URLs
# 20:33 aaronpk yeah so far nobody publishes pages with query strings
# 20:33 tantek and wait for someone to complain with a real world use case
# 20:33 Loqi aaronpk meant to say: yeah so far nobody publishes post pages with query strings
# 20:33 tantek no one says you have to accept webmentions from any type of URL
# 20:34 tantek use well design URLs or your webmention gets ignored :P
# 20:34 aaronpk yeah, they're called "query" strings after all... not perma-strings
# 20:37 XgF 410 Gone (your URL scared my webmention endpoint)
warden joined the channel
# 20:39 tantek that's too unforgiving. the webmention sender should be given a chance to clean up their source URL and try again. 410 implies don't webmention me ever again.
# 20:40 tantek barnabywalters - actually - there's 451 which is similar - Unavailable For Legal Reasons
# 20:41 tantek but what we want to communicate is not that the requested URL is unavailable for legal reasons, but rather that we can't parse your source URL for legal reasons
# 20:42 tantek so for something like webmention where the server is expected to go retrieve a URL provided by the client
# 20:43 tantek 452 Source URL unparsable for legal reasons - e.g. schemaorg patent policy does not give patent permissions to parse schema (only to publish)
# 20:44 tantek wow Google is filtering search results for schema within my website. I smell bad code or conspiracy
# 20:45 barnabywalters tantek: or filter bubbling? I see plenty of results for site:tantek.com schema
# 20:46 tantek barnabywalters - try the search box on tantek.com which includes extra parameters for strict date-time ordering of results
# 20:46 tantek none of my schema-related osfw3c posts show up
# 20:46 tantek I have to browse/nav to them explicitly by date
paulcp joined the channel
# 20:47 tantek 3 months and they haven't updated their TOS like they claimed they would - so I don't believe/trust anyone at Google about schema any more.
# 20:54 barnabywalters does wordpress automatically include a better canonical URL even if the page is accessed using that URL, or do you need a plugin for that?
# 20:54 KevinMarks_ I fought this with rel=tag, fairly succesfully (though now it's used as reason for not using rel=tag)
# 20:56 KevinMarks_ they used to use tag links with ? params too, which we policed out as an antipattern with technorati indexing
# 21:41 snarfed barnabywalters: once it's published publicly, yes, and automatically redirects to it
andreypopp, mko, tantek, pfefferle and bnvk joined the channel
josephboyle, ryana, paulcp, caseorganic, caseorga_, tantek and abrereton joined the channel