4 ms·

This was driving me mad for ages. I was blaming Twitter and my connection until I finally discovered it was Safari. This was the last straw for me. Having Safar

This was driving me mad for ages. I was blaming Twitter and my connection until I finally discovered it was Safari. This was the last straw for me. Having Safari on my desktop synced with my iPhone is so useful. Safari credit card and password storage works better than Chromes (imo) and I find it superior in almost every other way. But with todays issue, the t.co issue, and another issue where my entire Mac would randomly freeze and require a forced reboot on some videos (not sure if it was flash or HTML5) I've just had to give up and switch to Chrome.

untog 11y ago
They both show worrying things to me: - app crashing because a remote server is down? Bad coding. Bad QA. What else is lurking in that codebase? - weird bug affecting one domain? Fix it. Twitter is not a small site. But it's been months and they still haven't fixed it.
- coldtea 11y ago
  >app crashing because a remote server is down? Bad coding. Bad QA. What else is lurking in that codebase? That happens all the time with all kinds of apps, which nethertheless might be indispensable.
  - untog 11y ago
    If only it were open source, then we could check.
    - coldtea 11y ago
      If only it were open source and its users had both the knowledge required, the time AND cared enough to check.
theinternetman 11y ago
What's so special about t.co? No other sites are having a problem. Sounds like Twitters problem.
- untog 11y ago
  If you read the article you'll see an Apple engineer confirm it is a problem with Apple's code.
  - acdha 11y ago
    They confirmed that they're working on it but it's not clear that it was a purely a problem with Apple's code. Given all of the well-documented general stability problems t.co has had over the years it would not surprise me to learn that Chrome has a more aggressive recovery mode from a bad DNS server, failures returned by only one of the multiple servers which that hostname resolves to, etc. I've seen plenty of t.co errors in Firefox as well as Safari, but never in Chrome and the network stack is one of the areas where they've clearly put a lot of effort.
    - culturestate 11y ago
      When Hatcher said it was "lower level than WebKit," I took that to mean it has something to do with iOS/OS X's built-in Twitter integration.
      - influx 11y ago
        Seems like it could be DNS resolvers as well...
        acdha 11y ago
        That's what I was thinking as well: DNS, possibly something like the way their IPv6/4 reachability checks work (Happy Eyeballs is new enough to make that very easy to believe), etc.

untog 11y ago

They both show worrying things to me: - app crashing because a remote server is down? Bad coding. Bad QA. What else is lurking in that codebase? - weird bug affecting one domain? Fix it. Twitter is not a small site. But it's been months and they still haven't fixed it.

coldtea 11y ago
>app crashing because a remote server is down? Bad coding. Bad QA. What else is lurking in that codebase? That happens all the time with all kinds of apps, which nethertheless might be indispensable.
- untog 11y ago
  If only it were open source, then we could check.
  - coldtea 11y ago
    If only it were open source and its users had both the knowledge required, the time AND cared enough to check.

theinternetman 11y ago

What's so special about t.co? No other sites are having a problem. Sounds like Twitters problem.

untog 11y ago
If you read the article you'll see an Apple engineer confirm it is a problem with Apple's code.
- acdha 11y ago
  They confirmed that they're working on it but it's not clear that it was a purely a problem with Apple's code. Given all of the well-documented general stability problems t.co has had over the years it would not surprise me to learn that Chrome has a more aggressive recovery mode from a bad DNS server, failures returned by only one of the multiple servers which that hostname resolves to, etc. I've seen plenty of t.co errors in Firefox as well as Safari, but never in Chrome and the network stack is one of the areas where they've clearly put a lot of effort.
  - culturestate 11y ago
    When Hatcher said it was "lower level than WebKit," I took that to mean it has something to do with iOS/OS X's built-in Twitter integration.
    - influx 11y ago
      Seems like it could be DNS resolvers as well...
      - acdha 11y ago
        That's what I was thinking as well: DNS, possibly something like the way their IPv6/4 reachability checks work (Happy Eyeballs is new enough to make that very easy to believe), etc.