anarazel

35 ms·

421.
by anarazel 5y ago
For me this isn't an either or. I constantly use both together. For problems that a quickly and reliably reproducible I'll often just use the debugger (if rr is suitable, even better). But there's plenty problems that take a
422.
by anarazel 6y ago
Well, that could also be related to openbsd barely being used...
423.
by anarazel 6y ago
Anti-SLAPP statues aren't available everywhere in the US (nor I think I'm Canada). Importantly, there is none federally. Although sometimes state statutes can be used in federal court.
424.
by anarazel 6y ago
I found IWYU pretty annoying, due to its tendency to also include transitive includes. Some of those often end up about implementation details and are much more likely to be added/removed. But maybe the projects using it that I worked
425.
by anarazel 6y ago
I think it's a bit more complicated than that. Hard-to-work-with individual have been a problem in open source projects for a long time. What I think is changing is less the personalities of new contributors and more that [sub]projects
426.
by anarazel 6y ago
But only because perl was chosen as the default non-C language for postgres build system stuff maybe 20 years ago...
427.
by anarazel 6y ago
Sparc v9 had registers to change the consistency model. 1993. See "5.2.1.4 PSTATE_mem_model (MM)" in https://cr.yp.to/2005-590/sparcv9.pdf
428.
by anarazel 6y ago
Sparc did optional TSO 20+ years ago...
429.
by anarazel 6y ago
Database bulk load are where I have hit this many times. Often it's not viable to change the import to be a binary format (e.g cross databases, migrating from old proprietary software, etc).
430.
by anarazel 6y ago
Oh, aber not just memory pressure, a busy IO stack too. When there's multiple requests in flight already (concurrent processes, AIO, writeback) a part of the write may be issued as a separate (block layer) request, and IIRC there'
431.
by anarazel 6y ago
I see short writes pretty regularly under memory pressure. For writes well under 2GB. With io_uring there are extremely common (but got less so in very recent kernel versions). For different reasons for direct and buffered IO IIRC.
432.
by anarazel 6y ago
How can an individual understand and influence the risk preparedness of the whole grid? It's one thing to pay 3-4x for a day or be uncomfortably but safely warm. But a whole other thing to have costs increased by several orders of magn
433.
by anarazel 6y ago
> As a side note, don't do the following pattern: "BEGIN TRAN; INSERT INTO T ...; SELECT max(id) FROM T; COMMIT;". I used to do this, but this is a very bad habit that may be incorrect (assuming that id is an autoincrement
434.
by anarazel 6y ago
https://www.postgresql.org/docs/devel/storage-toast.html
435.
by anarazel 6y ago
Citus Data paid for one full-time Postgres developer for a while (me, before I left), and a few employees contributed things on and off. That didn't really provide Citus with a meaningful amount of control over the community, hence Mic
436.
by anarazel 6y ago
I agree that the whole PUBLIC thing is pretty confusing, and I doubt anybody would design it the way it is if we were start from scratch. Note that you can change the set of default privileges, including the grant to PUBLIC. E.g. ALTER DEFA
437.
by anarazel 6y ago
I wish some of these tests tested the performance of durable writes (via FUA writes or cache flush). Some otherwise fast drives are just ridiculously slow for those.
438.
by anarazel 6y ago
Ah, yes, shell history ftw. Of course you should ensure no filesystem is mounted or such: root@awork3:~# echo 4 > /sys/module/nvme/parameters/poll_queues root@awork3:~# echo 1 > /sys/block&
439.
by anarazel 6y ago
It's terribly documented :(. You need to set the nvme.poll_queues to the number of queues you want, before the disks are attached. I.e. either at boot, or you need to set the parameter and then cause the NVMe to be rescanned (you can d
440.
by anarazel 6y ago
> I tested various fio options, but didn't notice this one - I'll check it out! It might explain why I still kept seeing lots of interrupts raised even though I had enabled the I/O completion polling instead, with io_urin
441.
by anarazel 6y ago
Have you checked if using the fio options (--iodepth_batch_*) to batch submissions helps? Fio doesn't do that by default, and I found that that can be a significant benefit. Particularly submitting multiple up requests can amortize the
442.
by anarazel 6y ago
Plenty people spending a lot of their time working on Postgres have complained about Amazon's style of using open source. Among them myself (long before working for an AWS competitor). Most of us don't have problems, or even the o
443.
by anarazel 6y ago
I think you have to see that in the light of all the complaints about data loss when people moved from ext3 to ext4. Ted IIRC pushed back against that with arguments like yours, but eventually gave up (leading to a heuristic for detecting s
444.
by anarazel 6y ago
Not an iOS user, but I somehow expected rotation to work if I rotate two fingers against each other (I.e. draw a circle with two fingers at the same time). But I just need to swipe sideways with two fingers.
445.
by anarazel 6y ago
> Did they change libaio to work without O_DIRECT at some point? No, and I don't really forsee that happening at this point. > Or are you talking about io_uring for async file io? Yep. io_uring can do async buffered file IO. Ini
446.
by anarazel 6y ago
You can do network IO completion notification, instead of readiness notification with io_uring but not epoll. Which avoids the need for a separate syscall that needs to copy memory into the userspace buffers. That can be noticable. It shoul
447.
by anarazel 6y ago
> For those that don’t know, on Linux there is^H^H was really no such thing as properly async file system access (eg libc faked it in a similar fashion for aio) That's not quite right - there didn't use to be AIO for buffered
448.
by anarazel 6y ago
> I think I've only ever once seen any real use of comments in a real-life codebase. It was a postgres database, where columns that contained PII were commented according to whether it was plaintext or encrypted. ~10 years ago I wor
449.
by anarazel 6y ago
I'd like something roughly similar, to make the rr reverse debugger support io_uring. That likely can't work like most other syscalls, due to the memory only interface...
450.
by anarazel 6y ago
io_uring does allow to remove a lot of the syscall overhead, without polling. Many operations can be submitted with just one syscall. And ready completions can be consumed without a syscall at all. Additionally, compared to using epoll&#x2F

More ›