Twitter | Search | |
✕✕✕✕✕
The official Twitter account of Dwayne 'The Rock' Johnson
9,034
Tweets
249
Following
7,091
Followers
Tweets
✕✕✕✕✕ 14h
Replying to @_tessr @missionpie
it's egregious
Reply Retweet Like
✕✕✕✕✕ 14h
Replying to @smazero @missionpie
blocked
Reply Retweet Like
✕✕✕✕✕ 14h
a pie store that explicitly never sells key lime pie is so WILDLY incoherent to me that it's more easily understood as satire, an andy-kaufmann-esque pie shop as performative art (cough) (ahem)
Reply Retweet Like
✕✕✕✕✕ 14h
Replying to @rhyshiltner
Yeah, I guess my comment there is that As is an awkward name, it implies construction rather than introspection
Reply Retweet Like
✕✕✕✕✕ 14h
Replying to @terinjokes
right, but the metadata is canonically encoded in the Error() string, so it's not lost. it's not possible to program against it, sure, but my claim is you'd only ever do that in the immediate calling frame, anyway
Reply Retweet Like
✕✕✕✕✕ 14h
Replying to @StabbyCutyou
That feels right to me, too. But, unpopular opinion: I've never once needed to extract or inspect a callstack from an error?
Reply Retweet Like
✕✕✕✕✕ 15h
Replying to @terinjokes
If you need to get at the details of e.g. os.PathError, IME you always do it in the function that calls e.g. os.Chdir and can thus check for it directly, I've never seen a case where you need to get the details out of the middle of a wrapped error stack
Reply Retweet Like
✕✕✕✕✕ 15h
Replying to @peterbourgon
I haven't encountered an errors use case in the wild that wouldn't be solved by a package with three functions func Wrap(err error, context string) error func Wrapf(err error, context string, args ...interface{}) error func Contains(original error, target interface{}) bool
Reply Retweet Like
✕✕✕✕✕ 15h
I dislike the new/proposed , it's a large package surface area with lots of types and functions for esoteric use cases, feels inelegant, as if it were designed by a committee
Reply Retweet Like
✕✕✕✕✕ Mar 22
Replying to @dgryski @fatih @github
Reply Retweet Like
✕✕✕✕✕ Mar 22
great opportunity for an observability engineer in Berlin
Reply Retweet Like
✕✕✕✕✕ Mar 21
"HTTP/3? I thought we were still figuring out HTTP/2?" Explainer:
Reply Retweet Like
✕✕✕✕✕ Mar 21
Replying to @termie
I have two approaches, if it's a simple component I'll model it as a synchronous state machine, which is easy to test. If it's more complex, I send the input and use a test helper like this to check for expected output
Reply Retweet Like
✕✕✕✕✕ Mar 20
Replying to @peterbourgon
there’s not really a comparable situation in distsys, almost no “result” comes with an implementation that even works in a lab setting, much less one that can actually be deployed, or adapted and deployed
Reply Retweet Like
✕✕✕✕✕ Mar 20
i had this thought today as I read the HyperLogLog++ paper, it was so easy for them to like.. take the original result, optimizr with math, implement, test, and deliver something practical and useful—IMO because it all works on one machine
Reply Retweet Like
✕✕✕✕✕ Mar 20
Replying to @apenwarr
I don't quite follow the reasoning here
Reply Retweet Like
✕✕✕✕✕ Mar 20
Replying to @peterbourgon
i find this frustrating because while that ought to have transitive effects on theory, it often doesn't—probably due to the malformed incentives in academia (publish novelty or die), maybe also lack of real-world experience among authors??
Reply Retweet Like
✕✕✕✕✕ Mar 20
I think distsys is unlike a lot of other CS research topics because engineering/operational concerns actually dominate design
Reply Retweet Like
✕✕✕✕✕ Mar 19
Replying to @peterbourgon
it's the first time I've tried to boot a computer off a USB and got some security or permissions error, the future is stupid
Reply Retweet Like
✕✕✕✕✕ Mar 19
random question, but maybe somebody's done it: I have this AWS DeepLens sitting around and was unsuccessful in just putting like regular Debian on it so it can be a dumb server, has anyone hacked around on one of these things?
Reply Retweet Like