Conversation

Notices

1. Embed this notice
   Justine Smithies (justine@snac.smithies.me.uk)'s status on Sunday, 15-Dec-2024 22:54:07 JST Justine Smithies

   • Paul Martin
   @nowster@fedi.nowster.me.uk Welcome to #snac 😉
   
   • Embed this notice
     The Real Grunfink (grunfink@comam.es)'s status on Sunday, 15-Dec-2024 22:53:52 JST The Real Grunfink

     in reply to

     • Pachli
     • Paul Martin
     Hi, #snac author here. Thanks for reporting on the 'bot' account bug, I have to take a look at it.
     
     Regarding the other issue, snac do support Mastodon API pagination. Many apps do this correctly, like Tusky, Fedilab or semaphore.social, so you can travel the full timeline. Other ones only request the first page, I don't know why.
     
     CC: @nowster@fedi.nowster.me.uk @justine@snac.smithies.me.uk
     
   • Embed this notice
     Pachli (pachli@mastodon.social)'s status on Sunday, 15-Dec-2024 22:53:53 JST Pachli

     in reply to

     • Paul Martin

     @nowster @justine That's odd. Would you mind creating a test account on your server for me and sending the credentials to team@pachli.app. Then I can diagnose this further.

   • Embed this notice
     Pachli (pachli@mastodon.social)'s status on Sunday, 15-Dec-2024 22:53:53 JST Pachli

     in reply to

     • Paul Martin

     @nowster @justine Thanks for the test account.

     There are at least two #snac bugs.

     I've filed one of them at https://codeberg.org/grunfink/snac2/issues/241 (this also found a Pachli bug where errors during login weren't showing the underlying login error; I've fixed that now, thanks).

     The other is the existing https://codeberg.org/grunfink/snac2/issues/164 -- snac doesn't implement https://docs.joinmastodon.org/api/guidelines/#pagination so Mastodon clients can't iterate through timelines to load them a page at a time.

   • Embed this notice
     Paul Martin (nowster@fedi.nowster.me.uk)'s status on Sunday, 15-Dec-2024 22:54:06 JST Paul Martin

     in reply to
     @justine@snac.smithies.me.uk I notice that the current version of #pachli doesn't stop requesting the timeline when it refreshes. What I'll miss when I eventually migrate will be following hashtags.
     
   • Embed this notice
     The Real Grunfink (grunfink@comam.es)'s status on Sunday, 15-Dec-2024 23:57:39 JST The Real Grunfink

     in reply to

     • Pachli
     • Paul Martin
     Hi, #snac author here again. I've installed Pachli and I'm experimenting the same issue than @nowster@fedi.nowster.me.uk: it keeps requesting the first page over and over again, with no max_id nor since_id fields to paginate.
     
     CC: @justine@snac.smithies.me.uk
     
   • Embed this notice
     Pachli (pachli@mastodon.social)'s status on Monday, 16-Dec-2024 04:22:13 JST Pachli

     in reply to

     • The Real Grunfink
     • Paul Martin

     @grunfink @nowster @justine The link header is not being sent by snac.

     I have a test account with 4 statuses in the home timeline.

     /api/v1/timelines/home returns all 4 statuses and no link header in the response.

     /api/v1/timelines/home?limit=1 returns an array of 1 status, as expected, but is also missing the link header, so can't be paginated.

   • Embed this notice
     The Real Grunfink (grunfink@comam.es)'s status on Monday, 16-Dec-2024 04:22:13 JST The Real Grunfink

     in reply to

     • Pachli
     • Paul Martin
     Many apps work ok without the Link header kludge, they just keep track of the last seen post Id, and when they need more posts to be shown, they just use a since_id argument for the next page.
     
     As you say, the Mastodon documentation is a bit ambiguous.
     
     Anyway, I'll work on implementing it. Thanks for the conversation.
     
     CC: @nowster@fedi.nowster.me.uk @justine@snac.smithies.me.uk
     
   • Embed this notice
     knapjack (knapjack@elsewhere.cozysumo.space)'s status on Monday, 16-Dec-2024 06:50:49 JST knapjack

     in reply to

     • The Real Grunfink
     • Pachli
     • Paul Martin
     @grunfink@comam.es @pachli@mastodon.social @nowster@fedi.nowster.me.uk @justine@snac.smithies.me.uk Funny, it was on my to-do to investigate this. Tusky will retrieve successive pages (manually, not automatically) but Elk and the Tuba app will not. I suspect fix one, fix all.
     
     The Real Grunfink likes this.
   • Embed this notice
     The Real Grunfink (grunfink@comam.es)'s status on Monday, 16-Dec-2024 07:07:03 JST The Real Grunfink

     in reply to

     • Pachli
     • knapjack
     • Paul Martin
     Hi. I've just implemented the link header and Pachli seems to page correctly and not get stuck requesting the first page.
     
     This will probably also improve snac connectivity with other apps.
     
     CC: @nowster@fedi.nowster.me.uk @justine@snac.smithies.me.uk @knapjack@elsewhere.cozysumo.space
     
   • Embed this notice
     Pachli (pachli@mastodon.social)'s status on Monday, 16-Dec-2024 07:07:04 JST Pachli

     in reply to

     • The Real Grunfink
     • Paul Martin

     @grunfink @nowster @justine The Mastodon docs are a bit ambiguous about this. Arguably, if the link header is missing a client could assume it could behave as though it did exist, with the IDs of the first and last items in the returned page.

     That might be what snac is doing. For interoperability purposes it's probably better to always return the link header.