Hi, This seems like a logical feature to enable. Are there any downsides to enabling this?
Regards, Sander November 20, 2024 at 10:39 PM, "Willy Tarreau" <w...@1wt.eu mailto:w...@1wt.eu?to=%22Willy%20Tarreau%22%20%3Cw%401wt.eu%3E > wrote: > > Hi Lukas! > > On Wed, Nov 20, 2024 at 09:17:57PM +0100, Lukas Tribus wrote: > > > > > Hello! > > > > > > uri normalization is in experimental status since haproxy 2.4: > > > > http://docs.haproxy.org/3.0/configuration.html#4.4-normalize-uri > > > > URI normalization in HAProxy 2.4 is currently available as an experimental > > technical preview. As such, it requires the global directive > > 'expose-experimental-directives' first to be able to invoke it. > > > > > > Is it time to drop the "experimental" flag and consider this fully > > supported feature? Any H2 / H3 specific issues? Considering that it > > only touches the URI I would guess there are no issues. > > > Good point, I almost forgot about this! > > Some of the difficulties are that it's a set of enable/disable on the > URI itself but does not permit to perform them in a certain order, > nor does it permit to perform the operation on another header nor > a variable, nor even a location header. > > We've started to study the ability to do that using converters so > that users could chain them in any order, decide to decode before > checking or condition some tests to certain methods/uris etc, but > that required to implement the support for error reporting in sample > expressions from converters. Christopher had started digging around > this a while ago, and this was put to pause with the steady flow of > bug reports lately :-/ > > I think that once we have the ability to make converters cause the > processing to fail, then porting Tim's checks to converters is no > brainer and would offer more flexibility. In this case I'm not sure > there would still be a point in keeping the current configuration > mechanism. > > As much as I hate keeping experimental statuses for too long, I think > it would be reasonable to keep it one more cycle to give us a chance > again to address this globally. > > Thanks! > Willy >
