halayli

88 ms·

901.
by halayli 15y ago
Are you competitive?
902.
by halayli 15y ago
You are looking at techniques to merge 2 dictionaries into a new one. Some will find it useful.
903.
by halayli 15y ago
You cannot have a sane ORM that works with nosql and sql databases without compromising features & complexity. I am glad sqlalchemy deals with sql-only databases. You don't need an ORM for nosql. A library suffice.
904.
by halayli 15y ago
I wish Django uses sqlalchemy as the default ORM instead of inventing a squared wheel.
905.
by halayli 15y ago
I really like this book, it start with the very basics: Discrete Mathematics with Applications ( http://www.amazon.com/gp/product/0495391328/ref=oh_o04_s00_i... )
906.
by halayli 15y ago
first world problem?
907.
by halayli 15y ago
Consider using lthread for your c version. You'll save time and gain performance. ( http://github.com/halayli/lthread )
908.
by halayli 15y ago
I think using a generic OS for IO operations is not a good idea in general. We all do it but there got to be a better solution. We don't need kernel/userland isolation that introduces mode switching, stack copying, copying arguments etc...
909.
by halayli 15y ago
nginx saturates at ~18k req/sec / core with a latency of ~24ms. This saturation is not coming from nginx in particular but from OS limits (mode switching, stack copying etc.. for read/write system calls). There is nothing new in "modern HTT
910.
by halayli 15y ago
The OP easily dismissed numpy in Python and coroutines (e.g gevent)
911.
by halayli 15y ago
Things I can think about: - low level memory management - faster execution (in most cases) - cache line optimization - avoiding language implementation magic
912.
by halayli 15y ago
If you are checking your parameter types then you haven't used python enough.
913.
by halayli 15y ago
Better called, evil control structures. These macros add little value and makes it harder to understand the code.
914.
by halayli 15y ago
Why are you assuming it's linear?
915.
by halayli 15y ago
Tweaking variables.less is enough to make the site look unique. Someone should write a pseudo-random variables.less generator.
916.
by halayli 15y ago
It looks like a carbon copy of Apple App store / iTunes.
917.
by halayli 15y ago
Correct. The local variables address change, but you can still access them, and pass them to functions. What you cannot do is save a pointer of a variable and access inside begin()/end(). I thought I added a warning in the lthread_compute_b
918.
by halayli 15y ago
This has nothing to do with the kernel scheduler. lthread is a coroutine library, you can think of it as a micro task scheduler inside the process (userland). It's ideal for socket programming because it avoids using callbacks and minimizes
919.
by halayli 15y ago
Yes a coroutine user has to be aware that allocating on the stack has a penalty, similar to being aware that you cannot make a blocking call in an IO loop for example. On average, yielding ~10 calls deep results in copying ~75 to 100 bytes
920.
by halayli 15y ago
I am not married to the license. I noticed several complaints, so I am going to reconsider it. :)
921.
by halayli 15y ago
If you don't have large variables allocated on the stack, the copy is quite fast. This is easily avoidable by allocating on the heap.
922.
by halayli 15y ago
I do have few small code snippets + examples in the docs. https://github.com/halayli/lthread/tree/master/src/examples thanks for the heads up. :)
923.
by halayli 15y ago
I have a requirement that I couldn't get rid of yet, which requires me to know which scheduler it is going to run on before I let go of it. Once I manage to find a way around it I'll move to a global queue model.
924.
by halayli 15y ago
What happens is that it will block the other lthreads, bringing down the RPS to its knees. lthreads are simply coroutines and cooperation/trust are required to maintain fairness.
925.
by halayli 15y ago
It will not create a new pthread in the local scheduler but a local lthread scheduler gets created in the thread context. So if you want to create more than one lthread scheduler, you just have to create a pthread first and the new lthreads
926.
by halayli 15y ago
I am against IO event loops :). Scroll down to the bottom of the README and look for fibonacci(35) to see how I solve the problem mentioned in the link. The example is a naive HTTP server that computes fibonacci on every request and replies
927.
by halayli 15y ago
Thanks!
928.
Lthread - C coroutine lib with multicore support(github.com)
113 points by halayli 15y ago|73 comments
929.
by halayli 15y ago
It takes < 10s to reboot osx lion on my end (mba, iMac). My iMac was frequently crashing but it turned out to be from the RAM, Apple changed it and I never saw a crash again. Alas, the title is misleading.
930.
by halayli 15y ago
Why all the fuss? Email twitter and resolve it with them. Why everyone has to know about such a minor event?

More ›