#microformats 2015-05-08

2015-05-08 UTC
#
tantek
tommorris: right at the top: "Use case: IndieWeb "deleted" indication inline in content for static file services that don't support HTTP return codes. "
#
tommorris
yeah, but why just the http-equivs. you'd need to have both to discern anything useful
#
tommorris
I'm just not sure why you'd want one but not the other
#
tommorris
meta http-equiv is there for when you can't set headers. but headers are still sent - the meta http-equiv just overrides them in the document
#
tommorris
getting the overrides without the values they are overriding is a bit weird
tantek joined the channel
#
tantek
tommorris: re: "when you can't set headers" - exactly: "static file services that don't support HTTP return codes"
#
tantek
re: "you'd need to have both to discern anything useful" - not true - and certainly not true in this specific use-case where you do check for a 410 status code first, and then go parse the file.
#
tantek
point is to *not* have to *separately* parse for http equiv status code 410, but be able to just always run the microformats parser on the HTML and do *zero* parsing in your own code
#
tantek
re: "why you'd want one but not the other" - because this use-case is in the context of already having gotten the headers via curl
#
tommorris
but in the context of building the library, if I have access to both why not merge the two and return a combined dictionary of both headers and meta http-equivs
#
tommorris
difference between...
#
tommorris
A: MUST return http-equiv only
#
tantek
because it's extra work for the library
#
tommorris
B: MUST return http-equiv only and MAY return other headers parsed from the response and combine them together
#
tantek
and people already have header parsing / handling code
#
tantek
A is the minimum viable feature to satisfy the use-case - that's why
#
tommorris
should a microformats parser do anything with any of the other headers? behave like a proper HTTP user agent perhaps?
#
tommorris
if we're handling headers (and meta http-equiv are basically headers), we should probably follow what the headers tell us to do broadly in the same way browsers do
#
tantek
again - no need for the mvp for the use-case
#
tantek
less code = less chance of introducing bugs
#
tantek
a microformats parser may just get a URL (context) and a string of the curled data - e.g. that's how phpmf2 works now
elux_, fuzzyhorns, KartikPrabhu, tantek, KevinMarks_, eschnou, kez, gRegorLove, KevinMarks__, KevinMarks___, chiui and Left_Turn joined the channel
#
justgerrardz
edited /process-fr (+232) "/* En rapport */"
(view diff)
eschnou, adactio, kez, uf-wiki-visitor, elux, KevinMarks_, TallTed, KevinMarks__, KartikPrabhu, gRegorLove, bengo, chiui, KevinMarks___, fuzzyhorns and tantek joined the channel