#dev 2023-10-29

2023-10-29 UTC
gxt, alephalpha0, gerben, geoffo, [0x3b0b], eitilt, IWSlackGateway, kleb and rocto_ joined the channel
#
Zegnat
rubenwardy: always fun to read about switching experiences!
eitilt1, sgreger and [tantek] joined the channel
#
[tantek]
rubenwardy++ for a great migration write up!
#
Loqi
rubenwardy has 8 karma over the last year
bjoern, Sophie, [calumryan], jkphl, Sonja and [jeremycherfas] joined the channel
#
sknebel
Saphire: you serve a different cert through IPv6 on that domain, maybe that'S the issue? https://www.ssllabs.com/ssltest/analyze.html?d=auth.translunar.systems&hideResults=on&ignoreMismatch=on&latest
dervondenbergen joined the channel
#
Saphire
Uhh
#
Saphire
Oh, hmm, guess I have messed up record huh
#
Saphire
Thanks, that helps
#
Saphire
I'll need to chew on the dns records later
[sebsel] joined the channel
#
sknebel
hah, found some ticketing code in my site that I didnt remember adding. apparently I sent some tickets to GWG at some point :D
#
sgreger
Are there any opinions/precedence re respecting robots.txt when polling several thousand URLs (from my own content) to check for their existence?
[bjoern] joined the channel
#
sgreger
I'm testing a link checker tool to validate outgoing links on my old content and getting a huge amount of "Access denied by robots.txt". Feels right to obey, but also renders this task almost pointless.
#
Saphire
sgreger: how much of it is thanks to ad farms?
#
sknebel
sgreger: I feel like a linkchecker you individually run should not respect robots.txt
oodani joined the channel
#
Saphire
^
#
Saphire
Robots.txt is more for automated scrapping and aggregation
#
sgreger
Saphire: I concur. Presumably, most robots.txt excludes are in place to discourage other actors than genuine link verification.
#
Saphire
It might be useful as a signal "that might not be a webpage you are looking for" though
#
sgreger
sknebel: It's a bit of a conundrum. Technically it's a bot, but what it does is akin of me going through old links for both side's benefit.
#
sgreger
This entire automated validation of link-rot is a bit of a rabbit hole, I just realize. A lot of "Schrodinger's URLs": timeouts, rate limits, potentially temporary certificate issues, HTTP200s but with totally different owners/content, etc.
#
sknebel
something like "does the response still contain the title text" might be useful
#
Saphire
sknebel: I suspect a lot of "has a robots txt every covering this link" is just shitty ad plugs
#
Saphire
You know the kind, probably
#
sgreger
This already surfaced in yesterday's link-rot session: HTTP response alone is too weak a signal to determine live/dead/zombie.
#
Saphire
Yup..
tom, nertzy and barnaby joined the channel
#
GWG
Zegnat: Can I pick your brain on a question that we discussed in 2021-ish?
#
GWG
Zegnat: Specifically this one: https://github.com/indieweb/indieauth/issues/127 But there is a broader IndieAuth question...should the issuer URL have the metadata header on it...so you could find the metadata endpoint when looking at the issuer URL? Mine doesn't.
#
Loqi
[preview] [dshanske] #127 Where does the ticket endpoint go looking for the token endpoint?
IWSlackGateway and [tantek] joined the channel
#
Zegnat
I think I will need some time to read up, and I am currently with my head in a different subject at create day, so not sure I can actually give you anything smart right now
[KevinMarks] joined the channel
#
[KevinMarks]
"Does the response still contain the text I quoted? " would maybe be a better test
[calumryan], eitilt and gerben joined the channel
#
[tantek]
!tell [snarfed] BridgyFed docs PR to add 'hidden' attribute to markup examples to be kinder to screen readers and keyboard nav users: https://github.com/snarfed/bridgy-fed/pull/701
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
[preview] [tantek] #701 hidden attr for BF links
alephalpha0 joined the channel
#
GWG
Zegnat: I wonder if I should link to my birthday post in my h-card.
jonnybarnes joined the channel
#
Zegnat
GWG: right you can link it as you have it as a post!
geoffo joined the channel
#
GWG
I'm wondering how much longer I have to support token_endpoint and authorization_endpoint, as opposed to indieauth-metadata.
#
Zegnat
You can be the first to lead by example. Do it and start filing issues to places that do not support it.
#
GWG
Zegnat: Leading by example is supporting it. I do
#
Zegnat
It is just easier to break things if you want to file issues
#
GWG
Zegnat: So, disable the old way for all WordPress sites is your suggestion?
#
sknebel
You could do different endpoint urls in the two places and show a prompt if a user uses an app that does the old way
#
sknebel
That'd be interesting to track progress
#
GWG
Requires some thought, I suppose.
neceve, Xe and [jacky] joined the channel
#
[jacky]
Hmm using hidden for links
#
[tantek]
For programmatic links, not for humans to click
geoffo, AramZS, angelo, nertzy and mustastum joined the channel