4 ms·
I wonder if as they started to make this change they started finding numerous areas of the code where devs went "eh, it'll never be more than 140 chars... that'
by swalsh 10y ago
I wonder if as they started to make this change they started finding numerous areas of the code where devs went "eh, it'll never be more than 140 chars... that's our thing".
- apetresc 10y agoI don't think this change invalidates it. What they're basically doing is moving the list of reply-to's out of the body of the text into another field, which is being shown differently in the UI. What remains in the text field is still <=140 characters.
- matt4077 10y agoThis reminded me of the surprisingly interesting chapter on character counting in their docs: https://dev.twitter.com/basics/counting-characters https://dev.twitter.com/basics/counting-characters Summary: the probably did find problems like that, but long before this change when they encountered different languages and character sets. I'm sure running into the int32 limit for object ids was more of a hassle.
- danso 10y agoSpeaking of interesting writeups, their writeup on Snowflake, the system they use to generate unique 64-bit integers for tweets is itself a great read: https://blog.twitter.com/2010/announcing-snowflake https://blog.twitter.com/2010/announcing-snowflake
- fizx 10y agoThere was some stuff in search at one point that stored the length of a tweet in a byte. Making DMs longer was a database limitation at one point. Today though, I wouldn't expect it.