You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
QUIC's congestion control is based on TCP NewReno {{?RFC6582}}.
The next sections below roughly describe the behavior of different phases of congestion control. More concrete pseudo-code is only given in the Appendix later on but it's not linked from the original section.
On first reading, I missed the Appendix and felt the text lacking for concrete enough explanations to implement the feature. The question is what to describe where and what the authoritative definitions should be. If you know the NewReno RFC by heart you can probably make the right inferences to figure out how everything should work in comparison but is that how it should be?
An alternative (more convenient for the reader / skimmer) approach could be to mention NewReno only in passing and then provide a complete and self-contained description of the algorithms in the congestion control section directly, maybe even inlining the pseudo-code. If that is deemed to verbose and would bore readers that know enough about those algorithms already, it would be good to provide some guidance how to read the algorithm and what the assumptions are in the introductory paragraph about congestion control.
The text was updated successfully, but these errors were encountered:
The intent is that the text is complete without reading the referenced RFCs or pseudocode, so I think this section needs more text and possibly some inlined pseudo-code.
ianswett
changed the title
Congestion control section in recovery migth be easier to understand if it was self-contained
Congestion control section in recovery should be self-contained
Oct 14, 2019
It starts like this:
The next sections below roughly describe the behavior of different phases of congestion control. More concrete pseudo-code is only given in the Appendix later on but it's not linked from the original section.
On first reading, I missed the Appendix and felt the text lacking for concrete enough explanations to implement the feature. The question is what to describe where and what the authoritative definitions should be. If you know the NewReno RFC by heart you can probably make the right inferences to figure out how everything should work in comparison but is that how it should be?
An alternative (more convenient for the reader / skimmer) approach could be to mention NewReno only in passing and then provide a complete and self-contained description of the algorithms in the congestion control section directly, maybe even inlining the pseudo-code. If that is deemed to verbose and would bore readers that know enough about those algorithms already, it would be good to provide some guidance how to read the algorithm and what the assumptions are in the introductory paragraph about congestion control.
The text was updated successfully, but these errors were encountered: