httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (Ralf S. Engelschall)
Subject Re: config/234: REMOTE_IDENT not always available to mod_rewrite (fwd)
Date Mon, 17 Mar 1997 07:16:09 GMT

In article <> you wrote:

> From the fingers of Ralf S. Engelschall flowed the following:
> >
> >>     So mod_rewrite will incur the overhead of RFC1413 processing even if
> >>     the server configuration doesn't call for it.  Or am I missing
> >>     something?
> >
> >Correct, but this lookup is only done if there is a rewrite rule which
> >explicitly asks for this value, i.e. it uses the %{REMOTE_IDENT} variable. I
> >think it makes sense that a (always explicit) usage of this variable leads to
> >an explicit lookup of the identification string over the net if _it is still
> >not known_. Only this way you can write rulesets which do what they should.
> >
> >Think is is ok this way because mod_rewrite will only do the lookups if the
> >user asks it to do it. 

>     You make a reasonable case, but there's still something about this
>     that bothers me.  For one thing, it will still call rfc1413() even
>     if there's an explicit "IdentityCheck Off" directive - which doesn't
>     strike me as very friendly.

Hmmmmm... yes, sure. Hmmm....

>     I feel uncomfortable with the idea of a module using server
>     functionality (particularly with the overhead of rfc1413()) which
>     a) has an explicit enabling directive, b) isn't enabled by default,
>     and c) is used even if the directive has DISabled it.

>     Other things which use the remote_logname (such as the logging) are
>     at the mercy of it having been enabled.  If the problem is that it's
>     enabled, but rfc1413() hasn't been called by the time you need it,
>     *that* sounds like what should be fixed.  Otherwise, you're giving a
>     patient with an head-cold a tracheostomy.

Yes, best would be to fix this problem in the code, i.e.  to find the reason
why do_rfc1413 is not set. I looked at the code, but could not see any reason
why it should be not set. So I turned around and thought "hhmmm... ok, its
only a little problem and this way it would be ok, too". But perhaps you are
right. It would be better if we could find the reason instead of fixing the
symptom. Any ideas from where the problem comes? I personally was unable to
reproduce the problem myself....

                                       Ralf S. Engelschall

View raw message