Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
pspeter3
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
18 ms
·
211.
▲
by
pspeter3
9y ago
The statement was for 2009 when we had ~5 employees. I'm not sure what the timeline would look like if we had chosen the path you suggested. It does seem likely that we could have avoided an incremental rewrite. As mentioned in the pos
212.
▲
by
pspeter3
9y ago
That definitely looks like a bug. Thanks for reporting!
213.
▲
by
pspeter3
9y ago
Sounds great and inline with our thoughts. I'd love to see a blog post with more detail when it works!
214.
▲
by
pspeter3
9y ago
Thanks for the advice. Do you agree a potential optimization would be using React component state instead of store state?
215.
▲
by
pspeter3
9y ago
Thanks for the detail! Your description makes me wonder if there is a lurking memory leak.
216.
▲
by
pspeter3
9y ago
At a high level, we try to collect real user metrics for everything we can. Interana helps us analyze that data in near real time. For more detail, I hope we can write a blog post soon to answer your question.
217.
▲
by
pspeter3
9y ago
We have not yet but it is the goal we're working towarss. Migrating our legacy Luna code to be compatible with "use strict" and not rely on implicit global variables has been a grueling endeavor. We're actively testing a
218.
▲
by
pspeter3
9y ago
Thanks for your kind words! It took a lot of effort but we're happy with the result.
219.
▲
by
pspeter3
9y ago
Thanks! Please feel free to email me at phips@asana.com, message me on Twitter, or let me know what works best for you.
220.
▲
by
pspeter3
9y ago
Thanks for the advice! We believe that users remember the slowest actions as well. Your strategy sounds like it will help identify the long tail of rare yet slow actions.
221.
▲
by
pspeter3
9y ago
By Asana, I just meant employee. More people intuitively understood our metrics when we switched our framing.
222.
▲
by
pspeter3
9y ago
We used to track percentiles but discovered Asanas struggled to empathize with them. We switched to defining semantic performance bands (eg < 100ms, < 300ms, < 1s, < 3s, < 5s, < 10s). For each action, we track the percenta
223.
▲
by
pspeter3
9y ago
Thanks! Our hope is that Luna2 enables developers to more efficiently improve the product. While performance is still a top priority, more engineers can now focus on the product experience.
224.
▲
by
pspeter3
9y ago
Thanks! We would have really appreciated a resource like that when we started the process. We wanted to share our story in case it is helpful to others.
225.
▲
by
pspeter3
9y ago
Thanks for pointing that out. Our belief is that the majority of in app actions are now fast but that there are still exceptions. We're constantly trying to find new exceptions and remove them. Hopefully this thread alerts us of new on
226.
▲
by
pspeter3
9y ago
Page load performance is still a top focus of the company. While page load has improved significantly over the last six months, we know it has a long way to go. Our two largest performance issues are initial HTML time and bundle evaluation
227.
▲
by
pspeter3
9y ago
I deeply respect Elm and think many other teams members do as well. The fractal architecture concept was one of the influences to how we write Luna2 code. My main concern with Elm for Asana is long term risk. Choosing Elm means choosing a f
228.
▲
by
pspeter3
9y ago
Thanks for letting us know. That sounds like an exceptional case from our metrics. Would you be willing to share your Asana user ID so we could learn more?
229.
▲
by
pspeter3
9y ago
This is already in production for most views. Can you elaborate on your issues?
230.
▲
by
pspeter3
9y ago
Hi, I'm Phips and I'm one of the engineers behind Luna2. Please let me know if you have any questions!
231.
▲
Treating performance as a product: The technical story of Asana’s rewrite
(blog.asana.com)
135 points
by
pspeter3
9y ago
|
119 comments
232.
▲
by
pspeter3
9y ago
What does EDA stand for?
233.
▲
by
pspeter3
9y ago
Is there any documentation on why to use a typed array
234.
▲
by
pspeter3
9y ago
Thanks, I'll have to check this out!
235.
▲
by
pspeter3
9y ago
Thanks, that makes a lot of sense. It sounds like the compiler is providing most of the benefit. Agreed about GC. It's a significant portion of time in Asana for large task lists.
236.
▲
by
pspeter3
9y ago
So developers mark it manually?
237.
▲
by
pspeter3
9y ago
How does Moon automatically detect static virtual nodes? What makes Moon faster than the alternatives? Finally, how does Moon avoid GC pressure with large trees?
238.
▲
by
pspeter3
9y ago
This is our first community update, let us know what you think!
239.
▲
Asana Engineering Community Update: August 21-25
(blog.asana.com)
1 points
by
pspeter3
9y ago
|
1 comments
240.
▲
Office UI Fabric
(dev.office.com)
3 points
by
pspeter3
9y ago
|
0 comments
More ›