Thanks for the change. The rand.seed is however already called in job.go, as it only needs to be seeded once per execution (or job like here). It's much faster that way instead of applying a new seed for every call of the helper function.

This PR could however add a comment pointing that out for all the future readers of the code (myself included). For reference, the seed is set here:

Actually I found this issue when running with "calibration-enable" during "several times" and it always gave me the error message with the same random string.... :-)

Current code-base gives:
$ ./ffuf -ac -w words.txt -u https://notexisting.com/FUZZ
Error in autocalibration, exiting: Get https://notexisting.com/adminXVlBzgbaiCMRAjWw/: connect: no route to host
$ ./ffuf -ac -w words.txt -u https://notexisting.com/FUZZ
Error in autocalibration, exiting: Get https://notexisting.com/adminXVlBzgbaiCMRAjWw/: connect: no route to host
$ ./ffuf -ac -w words.txt -u https://notexisting.com/FUZZ
Error in autocalibration, exiting: Get https://notexisting.com/adminXVlBzgbaiCMRAjWw/: connect: no route to host

Adding the seed as in the PR.
$ vi pkg/ffuf/util.go
$ go build
$ ./ffuf -ac -w words.txt -u https://notexisting.com/FUZZ
Error in autocalibration, exiting: Get https://notexisting.com/adminPFpLgfKsynySawSo/: connect: no route to host

Good catch! You are right.

The autocalibration happens before job.Start(). Perhaps a good place to do the random seed initialization would be

I did some profiling on the amount of seed updates done when using this code from the PR. The code only calls the "seed update inside the RandomString function" 4 times per execution of the ffuf command.

I think this current PR code will not harm performance and is cleaner to understand when the rand.Seed + rand.Intn are inside the same function block. :)

I agree that with the current functionality it's not big of an issue. However I think that seeding the random every time a random value is needed is kind of an antipattern, and would have to be changed around whenever the function gets more use. Because of this I think it'd be better to try and find the proper place to do the random seeding.

How about updating the rand.Seed after https://github.com/ffuf/ffuf/blob/master/pkg/ffuf/job.go#L333 ?

Is this good to go?

CHANGELOG.md updated.