Conversation
So its like "mostly faster" on http benchmarks. |
Updated pull request to be more consistent with newly added |
Removed all benchmark results, just realized that I was running them on CI server. |
Benchmark results (sorry difference is inversed):
|
@bnoordhuis @isaacs this |
Fresh benchmark results:
|
!state.bufferProcessing && | ||
state.buffer.length) | ||
clearBuffer(this, state); | ||
} else { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This error message is kind of weird. Also, why nextTick? Seems like it should emit right now. Maybe something like 'Cannot uncork, not corked'
?
@isaacs updated, following tips from our previous discussion. Please check and let me know if I forgot anything. |
Still a few nits to pick...
I'll add a patch tomorrow to help out with this. |
Top two patches here change the API to what we decided on the mailing list: https://github.com/isaacs/node/commits/writev It'd still be good to get rid of the variadic functions and long functions. It's better to be less dry than to have variable function signatures or functions that are too long, or return early. This is hot code, need to kiss up to V8. |
@isaacs thanks, I've merged your changes into my branch. However I took slightly different approach for |
Add Writev method to StreamWrap class for writing mixed array of strings and buffers. Expose this method for TCP class.
/cc @isaacs @bnoordhuis
Previous discussion: #5246