Christophe Raffalli
Christophe Raffalli
It's done, I sent him an email ...
Sorry, my test code above is buggy: I wrote %17g instead of %.17g in the latest loop. I agree that 0.10000000000000001 is enoying.
Is this kind of idea below completely stupid ? If is (surprisingly) only twice as slow as ".17g", it can be read back, does the best for decimal numbers. I...
By the way it calls in average 2.14 times float_of_string per call for random floats.
Jean Michel Muller pointed me to Gay's code here: https://github.com/jwiegley/gdtoa It is big, but well tested and robust. We may particularise the code for mode 0 "shortest output". which should...
On 22-07-20 02:48:28, Xavier Leroy wrote: > See Ryu, the best(?) known algorithm for float-string conversion: https:// > dl.acm.org/doi/10.1145/3296979.3192369 > > Here is an implementation: https://github.com/ulfjack/ryu . However, the result...
Hello, There is a short code that I tested that find an exact solution by trial and error. It is necessarily slow. I will try to find back this code......
Remark: trailer is not used not implemented by most browser. Still the empty trailer should be there.
With cohttp-eio, http is almost enough ... except for Cohttp.Path.resolve_local_file. It would be nice if Http had this. A solution would be to move that function (of the Path module)...
And you can assign me to this task (i you want), I will need the same test for simple_httpd like the one for my PR #50 which is not very...