Aurélien D.

I don't know if this could be the cause of the warning but index_hash value can range from 0 to NB_L_HASH_WORD included, whereas hash_word array can only contain up to...

There is at least another function in tools.c (hash_ipanon) where index_hash value could reach NB_L_HASH_WORD and `hash_word[index_hash]` is used to access data.

Thanks for the longer report, it confirms it. When index_hash == NB_L_HASH_WORD (here it's maxed out to 7), we're overrunning hash_word array by 1. Both hash_anon and hash_ipanon are concerned...

> OpenSSL server $ openssl s_server -accept 22400 -Verify 1 -cert -key I ran some tests with 2.7 and 2.6 using self signed ca+certificates, here are my observations: Adding `-CAfile...

As for why the error only appears within freebsd jail, perhaps the jail has no ipv6 address configured so `socket(AF_INET6, ...)` fails in your case. `domain` argument from `socket(domain, ...)`...

Regarding @nbari's original request, can't we already achieve this by using an intermediary set-var to store the custom key used for the stick-on directive? e.g.: ``` # balance source stick-table...

Thanks for the detailed report Christian First bad commit is c6a7138420807db7cf0450208a3102584d424103 I'll have a look

Ok so previously `last_ispace` was only set to 0 when data was generated for `%hr` and `%hs` (to name a few). Since the faulty commit it may be set even...

I'm not 100% happy with the proposed fix because either way (avoiding or not double spaces), it isn't possible to tell if the current node was skipped because nothing was...

This patch series implements the logic from the comment above and addresses both the double space emission issue and OPT_MANDATORY option handling for all nodes. I'm waiting for feedback before...