| Tweetovi |
|
Matt Fleming
@fleming_matt
|
9 h |
|
I think BPF is definitely a good alternative. Just need to wait for the ecosystem to catch up so that it’s available for all distributions.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
10 h |
|
|
||
|
Matt Fleming
@fleming_matt
|
14 h |
|
It's pretty handy for investigating apps when you're unfamiliar with how they work internally. Like, where does time go when running host(1)? Disk i/o! It also spawns 4 tasks pic.twitter.com/Iv1wtiVpi8
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
14 h |
|
I wrote a BPF tool: github.com/mfleming/perfo…
I always wished time(1) displayed useful statistics like off-cpu time and whether the program was blocked on i/o (or sleeping in userspace). schedtime gives you that info by inspecting callstacks to work out why your app blocked.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
18 h |
|
Yes, CLRS!
Yeah, I think dynamic typing is mainly to blame for that annoyance because it's not always easy to look at code and know what it's doing. But for small functions and obvious designs it's hard to beat IMHO.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
18 h |
|
One of the things I love about Python is how close to pseudocode it can be. Perfect for quickly implementing algorithms. pic.twitter.com/NJFcs6se2M
|
||
|
|
||
| Matt Fleming proslijedio/la je tweet | ||
|
Cindy Sridharan
@copyconstruct
|
23 h |
|
Gandalf: An Intelligent, End-To-End Analytics Service for Safe Deployment in Cloud-Scale Infrastructure
Fascinating paper from Microsoft Azure on their "anomaly detection" system used for deployments, which also explains why traditional approaches to anomaly detection are flawed pic.twitter.com/jGtp0WdCFn
|
||
|
|
||
| Matt Fleming proslijedio/la je tweet | ||
|
Daniel Vetter
@danvet
|
6. velj |
|
Dave Chinner on how to do great & useful review if you're not a domain expert in that specific area
lwn.net/ml/linux-fsdev…
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
Fresh videos! twitter.com/poplconf/statu…
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
|
||
| Matt Fleming proslijedio/la je tweet | ||
|
Geoff Langdale
@geofflangdale
|
5. velj |
|
We had a system built around a Rabin-Karp approach, and it was "strided" if the string was big enough. It was... OK. However, that's a *lot* of computation so you tend to lose in the average case.
My favorite approach in x86 was to use PEXT as sort of a super-hash function. ...
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
I am dying.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
That’s a helluva lot of golang awesomeness packed into 20 slides. twitter.com/rakyll/status/…
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
I’m hoping for interesting things in the Observability space. twitter.com/redmonk/status… pic.twitter.com/U0RTiutvOm
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
I’m going to send a note to the cs book author to request that he include this image alongside the original one.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
Close!
s;push;pump;
Regular expressions still have their use.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
Right. It’s the recursion that trips you up because finite automaton have a finite number of states. You’re fine as long as you don’t need to remember which input you’ve seen, e.g. when pairing parentheses.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
That’s pretty much the number one thing to computational theory teaches working programmers.
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
Reminder: this is why you can’t parse programming language constructs with regular expressions. pic.twitter.com/NhLRIOzWpY
|
||
|
|
||
|
Matt Fleming
@fleming_matt
|
5. velj |
|
Maybe they’re not aware of the cost? I included a static regex compile inside some go code the other day instead of lifting it into a global var 🤦♂️
|
||
|
|
||