On 05/29/2013 01:13 AM, Amos Jeffries wrote:
> On 29/05/2013 12:55 p.m., Alex Rousskov wrote:
>> ------------------------------------------------------------
>> revno: 12865
>> committer: Alex Rousskov <rouss...@measurement-factory.com>
>> branch nick: trunk
>> timestamp: Tue 2013-05-28 18:55:39 -0600
>> message:
>> Make GCC on CentOS 5.3 happier by removing an unused static function
>> parse_acl_access was unused when USE_HTTP_VIOLATIONS was not
>> defined.
>> modified:
>> src/cache_cf.cc
>
> There is something very wrong with this change. At least two directives
> which are not wrapped in USE_HTTP_VIOLATIONS are of the type acl_access
> and should be using that function.
Hi Amos,
You probably missed the fact that there were two parse_acl_access()
functions. Only one of them (the sometimes unused one) was removed. The
auto-generated one was not affected.
The fix made GCC on CentOS 5.3 happier AFAICT: The build fails at a
different, seemingly unrelated to ACLs, point now.
> If possible can we get those cf.data.pre types updated and the 'old'
> acl_access variables all converted to the ConfigParser/LegacyParser
> mechanism ASAP.
As discussed in a separate email when acknowledging the boolean ACLs
commit, I will perform a global rename if there are enough good reasons
for doing that. Since the above request was based on a false premise
that the changes broke something, let's revisit that before we
inconvenience many pending patches with widespread renaming and lose the
easily-visible difference between old access_list and old ACLList.
Thank you,
Alex.
