@quark@ferengi.one No client, those were created using date -Is
and emacs. Off course all is UTF-8 encoded, but now Apache also announces content-type: text/plain; charset=utf-8
@prologic@twtxt.net Are you sure? The avatar file announced in my twtxt.txt was never fetched. Only non existing default avatars were fetched.
if-modified-since
header set, not all servers honor that header. Most of all, yarn does not: https://git.mills.io/yarnsocial/yarn/issues/430
@movq@www.uninformativ.de What I would really like to see if jenny could use HTTP range requests to fetch only new content.
E.g. it could refetch only last twtext line of last request to make sure it starts off at correct position.
I guess there are twtxt files that only grow, then this will save a lot bandwidth over time.
For twtxt files that “forget” older content this situation would be detected and as a fallback the whole twtxt file could then be fetched.
@eldersnake@yarn.andrewjvpowell.com Maybe they are just lurking (and learning)?
;-)
@prologic@twtxt.net I would like to see “header” lines in twtxt.txt parsed.
Personally I started looking at some twtxt files with curl and saw information about avatar images.
I assumed that to be sort of standard and mentioned my avatar image in my stackeffect.txt. But it was not “avatar.png”.
Later I saw in logfiles that the info was totally ignored and instead several “avatar.png” locations were tried by the pulling side.
When information in “header” of twtxt file were respected one could easily change avatar file to one with a new filename and there would be no caching problem.