"Leslie P. Polzer" <[email protected]> writes: > Rayservers wrote: >> >> Just FYI, I stumbled across this today, its good to see Lisp doing so >> well:
Great, someone read my blog :-) >> http://john.freml.in/teepeedee2-release >> http://john.freml.in/ >> >> 10k requests per second in Lisp! >> >> Anyone have benchmarks with weblocks? > > Weblocks uses Hunchentoot underneath. Some random numbers: > > Hunchentoot 80reqs/s, Weblocks 40reqs/s. > > Note that these numbers hold for a full page request, not just a tiny > "get a snippet of text" AJAX request over a channel that is probably > already open. The tpd2 benchmark (11k requests/second) was for a full connection open and closing (not a persistent connection which tends to crash apachebench) with tpd2 on one core of my laptop and apachebench on the other core. Although the page was small, it was dynamic and a complete page. The tpd2 architecture means that you can put in whatever HTML you like and only pay in performance for the dynamic pieces, so a larger page should only make the network stuff slower. (At least, that's the theory. . .) > Additionally HT doesn't support epoll() or other fancy stuff like IOlib > does. (tpd2 doesn't use iolib but calls the libc syscall wrappers with its own CFFI definitions. The performance issues with iolib that required this have mostly been fixed I think, so it might be nice to move back. On the other hand, a more direct set of machine insns for getting to the kernel than going through libc would be very cool . . .) tpd2 has its own HTML generation that is IMHO much nicer that cl-who (attribute spell checking, less annoying syntax, etc.) and definitely much faster. It would be great to have weblocks running on tpd2 (for both projects), in my opinion, but actually that would require a lot of integration work that nobody is planning on prioritizing :-( Weblocks really needs someone to work full time on it to develop its great potential and mucking about with the low-level stuff isn't the first thing to do -- shame there's no funding for it, but I guess someone needs to invest the time in documenting it before the imaginary megacorps start contributing to the Weblocks Foundation(tm) ;-) [...] --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "weblocks" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/weblocks?hl=en -~----------~----~----~----~------~----~------~--~---
