fmaury/seirdy.one
1
0
Fork 0
mirror of https://git.sr.ht/~seirdy/seirdy.one synced 2024-11-27 14:12:09 +00:00
Code Issues Projects Releases Wiki Activity

Clarification around round-trip window sizes

Browse source
This commit is contained in:
Rohan Kumar 2022-09-22 10:51:53 -07:00
parent c2fafa8d97
commit f630c8b670
No known key found for this signature in database
GPG key ID: 1E892DB2A5F84479
2 changed files with 6 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
content/posts/website-best-practices.gmi
View file

@ -1806,7 +1806,8 @@ I opted to meet the criterion a different way. Since all my pages are linked by
⁶ uBlock Origin is a popular browser extension for content filtering; it's the most popular Firefox add-on. It includes a built-in feature to block all media elements exceeding a user-configurable size threshold.
⁷ If you'd like more details about TCP, here's a great introduction:
⁷ Technically, this is only true in idealized circumstances and if you have 0-RTT enabled. If your requests are not idempotent, you probably should stick with 1-RTT. A TCP handshake actually involves one additional round-trip, so you could probably start off with a 30-kilobyte window on HTTP/2 with 0-RTT. HTTP/3 servers with 0-RTT do use a true "zero-round-trip" connection, but they use QUIC instead of TCP. QUIC uses similar sizing logic but I'm not sure about the exact values.
If you'd like more details about TCP, here's a great introduction:
=> https://hpbn.co/building-blocks-of-tcp/ High-Performance Browser Networking, by Ilya Grigorik
⁸ Iterating through a list of font names to see if each one is available on a user's system is a slow but effective way to determine installed fonts without being granted permission to use the Font Access API. BrowserLeaks has a demo of this approach.

4
content/posts/website-best-practices.md
View file

@ -1860,7 +1860,9 @@ A special thanks goes out to GothAlice for the questions she answered in <samp>#
[^5]: One example is a utility class for the `image-rendering` property. I use it on images that look better with `pixelated` rendering. This is a property of the image contents, not the image semantics or placement; a class makes sense.
[^6]: {{<mention-work itemprop="citation" itemtype="TechArticle">}}{{<cited-work name="High-Performance Browser Networking" url="https://hpbn.co/building-blocks-of-tcp/" extraName="headline">}} by {{<indieweb-person first-name="Ilya" last-name="Grigorik" url="https://www.igvita.com/" itemprop="author">}}{{</mention-work>}} gives a great introduction to how TCP works, if you'd like more details.
[^6]: Technically, this is only true in idealized circumstances and if you have 0-RTT enabled. If your requests are not idempotent, you probably should stick with 1-RTT. A TCP handshake actually involves one additional round-trip, so you could probably start off with a 30-kilobyte window on HTTP/2 with 0-RTT. HTTP/3 servers with 0-RTT do use a true "zero-round-trip" connection, but they use QUIC instead of TCP. QUIC uses similar sizing logic but I'm not sure about the exact values.
{{<mention-work itemprop="citation" itemtype="TechArticle">}}{{<cited-work name="High-Performance Browser Networking" url="https://hpbn.co/building-blocks-of-tcp/" extraName="headline">}} by {{<indieweb-person first-name="Ilya" last-name="Grigorik" url="https://www.igvita.com/" itemprop="author">}}{{</mention-work>}} gives a great introduction to how TCP works, if you'd like more details.
[^7]: This one-kilobyte limit is a semi-arbitrary rule-of-thumb I came up with. It's a simple number easier to work with than the number of bytes remaining in the earliest contentful round-trip, yet it typically falls within that quota.