Why offer an option to a directive which makes it behave differently AND have an explicit directive for that exact behaviour ? I see no gain from this but a lot of potential harm in terms of user confusion. Better be explicit and unambiguous -> 2)
On Mon, Dec 16, 2013 at 6:37 AM, Christophe JAILLET < christophe.jail...@wanadoo.fr> wrote: > Hi, > > I was about to commit a patch for mod_proxy.c, but around line 2230 > > > else if (!strcmp(cmd->path, "~")) { > > cmd->path = ap_getword_conf(cmd->pool, &arg); > > if (!cmd->path) > > return "<Proxy ~ > block must specify a path"; > > !cmd->path should be !cmd->path[0] because 'ap_getword_conf' does not > return NULL but an empty string, so the test can never be true. > > However, I've searched in the doc to get explanation of this '~' in <Proxy > > block. > There is nothing about it. > > Apparently <Proxy ~ xx> is equivalent to <ProxyMatch xx> > So, should we: > 1) just leave it as it is. > 2) remove this behavior, which has never been documented. > 3) update doc. > > > My choice is 2) because > - it has never been documented, so it is likely that no one use it > - apparently it is equivalent to <ProxyMatch > > > > Your opinion ? > > Best regards, > CJ > >