on the vsl format - VIP23 (VSL refactoring) design sketch

Dridi Boukelmoune dridi at varnish-software.com
Fri Apr 12 18:37:56 UTC 2019

On Fri, Apr 12, 2019 at 6:27 PM Nils Goroll <nils.goroll at uplex.de> wrote:
> On 12/04/2019 17:57, Dridi Boukelmoune wrote:
> > The prefix, which may be optional, and by nature is supposed to be the
> > first field, and coincidentally is of variable length, is a problem.
> I do not understand why. The VSL client code which matches the prefix can do so
> against the string field as before, it is just not the start of the record payload.

The prefix is too important as an optional field to move to the last
position and jeopardize its presence because of vsl_reclen.


More information about the varnish-dev mailing list