Two bots have found each other
I'm dying 🤣
......(▼ヘ▼#) انور
Sia✨
I want a bot too now
Sia✨
I'll not
Another one bites the dust...!
Banned 路亚十三.
Reason:
crowdfund add + non-english
labyrinth
I found a weird coredump to occur in C++. Can anyone help me understand what is going on?
https://pastebin.com/GYwbNPLa
basically, I am going to sort a 2D vector with my custom lambda, where the inner vector has a size of 2.
for vector<vector<int>> arr, if arr.size() is 2, the code will work. but if arr.size() is 10000, it will fail, the lambda will get int instead of vector<int>.
Chat Boss
labyrinth sent a code, it has been re-uploaded as a quote
why on earth there is an empty arg to the lambda
0x00000000005e2fc1 in mysort(std::vector<std::vector<int, std::allocator<int> >, std::allocator<std::vector<int, std::allocator<int> > > >&)::$_0::operator()(std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&) const (this=0x7ffffff12c6f, l=std::vector of length 0, capacity 0,
r=std::vector of length 2, capacity 2 = {...}) at merge-intervals.cpp:10
10 if (l[0] == r[0]) {
Ludovic 'Archivist'
I found a weird coredump to occur in C++. Can anyone help me understand what is going on?
https://pastebin.com/GYwbNPLa
basically, I am going to sort a 2D vector with my custom lambda, where the inner vector has a size of 2.
for vector<vector<int>> arr, if arr.size() is 2, the code will work. but if arr.size() is 10000, it will fail, the lambda will get int instead of vector<int>.
sort may never terminate here, you should sort on < not on <=
Sweet girl
Where are my united State guys
Sweet girl
Okay
Orange Juice
printf("Meow");
Suhrob
Std::cout << "hi guys";
Pavel
Actually, are printf and outputting into cout synchronized in any way?
Pavel
Or it could be printed in reverse order as well? I mean because cout has a buffer, I assume
Ludovic 'Archivist'
Actually, are printf and outputting into cout synchronized in any way?
Yes, you can desync them for better cout performance
Sweet girl
Where are my united State guys
Suka
Oliver
Where
Oliver
Can you help me
Orange Juice
hello world("printf"):
klimi
Why
uhm... idk... they decided that united states are in america... I am not the one who put them there
Sweet girl
Okay
Daniya
H
Moita
Seyi
Anyone have experience using Mozilla NSS
harmony5 🇺🇳 ⌤
Anyone have experience using Mozilla NSS
Don't ask meta questions. In other words, don't ask to ask. Questions like "Does anyone know XYZ?", "Has anyone used XYZ?" or "Can someone help me?" are all considered meta questions because they don't specify what your actual problem is.
These questions give the impression that you want people to approach you and offer their help as if they don't have any other work to do. Now doesn't that expectation make you look like an idiot?
If you have a question ask it directly. You are more likely to get a response that way.
ᗪ
Good evening
ᗪ
Kk byee
~
Hello i want to ask
I have 8 cores
Which design is better?
1. Design 1 :
Create 8 coroutine schedulers that handle http requests, and the database operations are also handled by these schedulers
2. Design 2 :
Create 7 coroutine schedulers that handle requests, and 1 scheduler dedicated to handle database operations communicating with postgres non blocking api
Ludovic 'Archivist'
Hello i want to ask
I have 8 cores
Which design is better?
1. Design 1 :
Create 8 coroutine schedulers that handle http requests, and the database operations are also handled by these schedulers
2. Design 2 :
Create 7 coroutine schedulers that handle requests, and 1 scheduler dedicated to handle database operations communicating with postgres non blocking api
I would look into what trantor does. Generally it uses 1 coroutine scheduler per core, and then an extra thread pool for database operations
~
I would look into what trantor does. Generally it uses 1 coroutine scheduler per core, and then an extra thread pool for database operations
Can you explain why i should use os thread pool for database operations, not coroutines?
Ludovic 'Archivist'
Can you explain why i should use os thread pool for database operations, not coroutines?
Because not all databases support async operations. It will allow you to switch databases on demand. Also many database connectors are synchronous and use TCP which is a terrible bottleneck in that it may, in worst case, on Linux, block a thread for up to 13 minutes. So thread pool for network resources is a disgraceful but more robust solution than just relying on coroutines
Ludovic 'Archivist'
(I want to mention that I hate that fact a lot)
Ludovic 'Archivist'
Such a network bug used to exist in Apache Zookeeper
~
Because not all databases support async operations. It will allow you to switch databases on demand. Also many database connectors are synchronous and use TCP which is a terrible bottleneck in that it may, in worst case, on Linux, block a thread for up to 13 minutes. So thread pool for network resources is a disgraceful but more robust solution than just relying on coroutines
At this my hobby project :) i want to just support postgres because its the most used open source db, if im not wrong postgres has a non blocking api that can be used to do queries asyncronously, what do you think about it?
Ludovic 'Archivist'
At this my hobby project :) i want to just support postgres because its the most used open source db, if im not wrong postgres has a non blocking api that can be used to do queries asyncronously, what do you think about it?
I have never used that API, for a hobby project it is probably fine to just use your normal coroutine runners for the DB stuff
Ludovic 'Archivist'
> Here the doc if you want to read https://www.postgresql.org/docs/current/libpq-async.html
Ludovic 'Archivist'
The bot removed your post because you joined recently
~
Hello i want to ask
I have 8 cores
Which design is better?
1. Design 1 :
Create 8 coroutine schedulers that handle http requests, and the database operations are also handled by these schedulers
2. Design 2 :
Create 7 coroutine schedulers that handle requests, and 1 scheduler dedicated to handle database operations communicating with postgres non blocking api
And then to my main question, which is better about that 2 designs?
Ludovic 'Archivist'
Hello i want to ask
I have 8 cores
Which design is better?
1. Design 1 :
Create 8 coroutine schedulers that handle http requests, and the database operations are also handled by these schedulers
2. Design 2 :
Create 7 coroutine schedulers that handle requests, and 1 scheduler dedicated to handle database operations communicating with postgres non blocking api
I would try 1. first, check if everything still works correctly under load (I do not know how threadsafe that postgres api is), if it works it works (I would still be wary, never trust concurrency), otherwise I would make dedicated database executor threads (similar to 2., you will want requests to be locked to a specific thread tho, to avoid weirdness)
Ludovic 'Archivist'
You will have a pool of connections anyway as the async API locks the entire connection too
Ludovic 'Archivist'
Hence why trantor/drogon use a threadpool anyway imho
Ludovic 'Archivist'
PQgetResult is synchronous anyway so that entire API looks entirely useless to me if what you want is improved throughput
Ludovic 'Archivist'
It is only good if you cannot hold the entire result set in RAM
Ludovic 'Archivist'
Keep in mind that large result sets in database software are pushed to disk so if you reach that point in terms of large queries you would need to question the sanity of doing processing away from the database
Ludovic 'Archivist'
rust-postgres doesn't use the libpq, they implement the whole postgres stack themselves
Ludovic 'Archivist'
This way they are not tributaries of the original driver's design
~
rust-postgres doesn't use the libpq, they implement the whole postgres stack themselves
Do you know what lib to use to achieve the same that can be used with independent coroutine library? I use libcoro library
Ludovic 'Archivist'
Do you know what lib to use to achieve the same that can be used with independent coroutine library? I use libcoro library
You would need to implement your own
Vlad
Do you know what lib to use to achieve the same that can be used with independent coroutine library? I use libcoro library
Do you have such strict requirements for your personal project that you need all of this?
Vlad
Or this is shiny object syndrome?
~
Do you have such strict requirements for your personal project that you need all of this?
Actually im trying to create backend framework 😅
Ludovic 'Archivist'
Actually im trying to create backend framework 😅
Then yeah, you would want to implement the postgresql protocol using libcoro yourself, it's not been done yet
Ludovic 'Archivist'
Good luck, welcome to hell
Ludovic 'Archivist'
Well, to be fair with libcoro you at least already have TLS out of the box
Vlad
Well, to be fair with libcoro you at least already have TLS out of the box
tls is absurdly annoying to implement and make work seamlessly. so that's definitely worth
Ludovic 'Archivist'
Vlad
I remember on my first internship the whole team was trying to deal with random tls socket connection drops for a week
Ludovic 'Archivist'
Ludovic 'Archivist'
TLS is hell
Ludovic 'Archivist'
Parsing is PostgreSQL on top of that is an entire extra flavour of hell
Reported T $$$ [7560409545] to admins.
Paulo
Guys, anyone has any tutorial or book to recommend about lock-free c++ programming ?
Ludovic 'Archivist'
Guys, anyone has any tutorial or book to recommend about lock-free c++ programming ?
Asynchronous and Lock-Free Programming in C++ published by Packt
Paulo
is this book?
https://www.packtpub.com/en-RO/product/c-reactive-programming-9781788629775/chapter/asynchronous-and-lock-free-programming-in-c-4/section/asynchronous-and-lock-free-programming-in-c-4?srsltid=AfmBOopJaHLQbS3FK0xvOu3c0p9aNbX8YilnOM2_Ee9yY4w8cj9FNEgP
Ludovic 'Archivist'
Wait, I got the title wrong, and it is not one book
Ludovic 'Archivist'
is this book?
https://www.packtpub.com/en-RO/product/c-reactive-programming-9781788629775/chapter/asynchronous-and-lock-free-programming-in-c-4/section/asynchronous-and-lock-free-programming-in-c-4?srsltid=AfmBOopJaHLQbS3FK0xvOu3c0p9aNbX8YilnOM2_Ee9yY4w8cj9FNEgP
This one is good, but there is also Asynchronous Programming with C++
Ludovic 'Archivist'
I also suggest you look for moodycamel 's lock-free queue designs
Ludovic 'Archivist'
https://moodycamel.com/blog/2014/a-fast-general-purpose-lock-free-queue-for-c++