On Thu, 2007-08-02 at 00:40 +0200, Henrik Nordstrom wrote: > On tor, 2007-08-02 at 01:32 +0300, Tsantilas Christos wrote: > > > At the end I convert Config.readAheadGap and Config.quickAbort.min/max > > to b_int64_t types. > > Ok. > > > But this is changes the default units for Config.quickAbort.min/max > > parameters and make them different than those of squid26 configuration > > file. Is it any problem with it? > > Not in my eyes. > > And certainly not if changing the code to no longer accept unitless > specifications. > > These directives have always been unit based, and at least 2.6 (probably > 2.5 as well) gives big fat warnings issued if the user specify them > without unit. So the change to make Squid-3 require an unit is minor. > > My view is that any Squid-2 configurations out in the field without an > unit specified is bad configurations and should be fixed, even for > Squid-2. So having Squid-3 reject them is actually an improvement. The > fixed configuration will work as expected in both versions.
If a unitless configuration is rejected, there is no problem. If it is accepted (with a warning) and then interpreted differently from Squid2, I think there is a problem because many admins will [continue to] ignore the warning. Can you make the units required in this particular case? If there is consensus that units should be required everywhere, I am fine with that (but still worried). Alex.
