Nils Goroll
Nils Goroll
bugwash: * pcre2 might offer a way for stack bounds checking * We want to switch to pcre2 #3559
Can anyone help me understand the motivation? In my mind, the time scale of outdated values for non volatile access should be irrelevant to vsc clients.
@asadsa92 thank you for the clarification and research. Within varnishd (at the write end for statistics), `volatile` kind-of makes sense: Because we update values (e.g. increase/decrese), outdated values cause inaccurate...
Does this PR still serve a purpose?
right after openening the ticket I had an idea leading further: ``` $ varnishlog -r complete_log -g session -q 'vxid == 58439167' * > 58439167 - Begin sess 0 PROXY...
bugwash: * Add callback to VSL API at every segment switch if transactions had been dropped * add `varnishlog` parameter to report drops every _x_ seconds (0 = off, default...
this is major release material due to the required api chance
while at it, we might also want to improve error reporting in general, see #3357 we could have pointers to varnish-faq(7) in confusing error messages
changes.rst which I had written before noticing the revert, dropping here for possible reuse: ```rst * Previously, when sending a backend request failed, varnish would always generate a 503 response...
FTR, before we actually do it, I'd like to check Solaris / SunCC. I fear it might still need `-mt`.