Hacker News new | past | comments | ask | show | jobs | submit login

It's more in the category of "the new kid on the block tries to implement to spec and discovers what everyone else already knows: spec's hosed."

I once received a bug report on a site that consistently went down after a computer woke from sleep... But only if the computer was a macintosh, and only if the browser was Chrome. It turned out that the root cause was that when the machine slept and reawoke, XML HTTP requests that were attached to timers in an open webpage would fire all at once.

On Windows and Linux, apparently, the network stack would dutifully pause those requests while the radio took a moment to reestablish connection. Mac OS x, adhering to the spec, did not pause but instead immediately reported on wake that the network was unavailable.

So, the other browsers on Mac OS wisely broke speck and ignored the first couple network down that came in after sleep, quietly retrying rpcs. Chrome adhered to spec and dutifully reported the dropped network as an error that failed all those rpcs.

As a result, client's page was broken, but only on Mac os, only on wi-fi, and only on Chrome. Would you guess that their first solution was to painstakingly rewrite all of their set timeout logic to move the retries up to the JavaScript layer, or would you guess that their first solution was to report a bug to Google and tell their regular users Chrome was broken?

In any case, it's a moot point now because at some point Chrome changed their network stack implementation to match everybody else's. ;)




It is std on std case ?




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: