[DiscordArchive] And how would multithreaded read or write work?
[DiscordArchive] And how would multithreaded read or write work?
Archived author: Deleted User • Posted: 2018-06-18T14:29:18.881000+00:00
Original source
And how would multithreaded read or write work?
Archived author: Quantam • Posted: 2018-06-18T14:29:32.597000+00:00
Original source
So you give it the end offset in the first call, then the start offset in the second call
Archived author: schlumpf • Posted: 2018-06-18T14:29:34.109000+00:00
Original source
In fact it is good that the kernel handles that for you rather than applications having to handle shared ownership between threads.
Archived author: Deleted User • Posted: 2018-06-18T14:30:01.737000+00:00
Original source
So you mean they should have only a readAt method
Archived author: schlumpf • Posted: 2018-06-18T14:30:14.025000+00:00
Original source
Quantam has a point that `read (pos, size) -> byte[]` is a better API than `seek& (pos); read& (size) -> byte[]`.
Archived author: Deleted User • Posted: 2018-06-18T14:30:45.566000+00:00
Original source
That’s alright
Archived author: Deleted User • Posted: 2018-06-18T14:31:06.840000+00:00
Original source
I thought you envisioned the Java’s no seek stream api
Archived author: Deleted User • Posted: 2018-06-18T14:31:17.792000+00:00
Original source
You cannot reposition streams
Archived author: Quantam • Posted: 2018-06-18T14:31:41.926000+00:00
Original source
Having separate handles for every thread is a bunch of extra kernel objects that are entirely unnecessary
Archived author: Quantam • Posted: 2018-06-18T14:32:08.038000+00:00
Original source
And falls over and dies if you're doing async I/O with a ton of queued requests