Jame,
In // you start all requests in // as a
result it is much faster.
50 request on http://example.com
- synchronous 8.51s
- asynchronous 0.19s
The test is very simple to do.
- asynchronous ./build/curl-http -v
-a [multi url] >/dev/null
- synchronous ./build/curl-http -v
-s [multi url] >/dev/null
Note: I still have an issue with huge
URL (1G) fail both in both mode. As it fail with standard libCurl
it should come from a mistake on my side in initializing curl
easy.
Fulup
ps: any idea on why it fail on huge URL
is welcome
https://upload.wikimedia.org/wikipedia/commons/d/dd/Big_%26_Small_Pumkins.JPG
On 09/02/2021 15:57, James Read wrote:
On Tue, Feb 9, 2021 at 12:56 PM Fulup Ar Foll via curl-library <curl-library@cool.haxx.se> wrote:
Hi,
I posted on github a template for libcurl asynchronous usage with
libuv+libsystemd mainloop abstraction:
https://github.com/fulup-bzh/libcurl-mainloop .
While I spend in significant time searched documentation and web, I best
case found partial information and worse case invalid model. Looking to
the number of lost people I found with equivalent requirement, it is
clear that I'm not the only one searching for a proper libcurl
asynchronous model.
Have you tested it with a sizeable number of URLs? What kind of throughput do you get?
James
Hopefully my code may help the community. If you think ot might be
useful feel to use/modify it as you like.
Fulup
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette: https://curl.se/mail/etiquette.html
------------------------------------------------------------------- Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.html
