From apwww@hyperreal.org Sun Aug 17 04:48:31 1997 Received: (from apwww@localhost) by hyperreal.org (8.8.5/8.8.5) id EAA11617; Sun, 17 Aug 1997 04:48:31 -0700 (PDT) Message-Id: <199708171148.EAA11617@hyperreal.org> Date: Sun, 17 Aug 1997 04:48:31 -0700 (PDT) From: Dean Gaudet Reply-To: dgaudet@apache.org To: apbugs@hyperreal.org Subject: request_config field in request_rec is moderately bogus X-Send-Pr-Version: 3.2 >Number: 1004 >Category: apache-api >Synopsis: request_config field in request_rec is moderately bogus >Confidential: no >Severity: non-critical >Priority: medium >Responsible: apache >State: suspended >Class: change-request >Submitter-Id: apache >Arrival-Date: Sun Aug 17 04:50:01 1997 >Last-Modified: Mon Feb 23 03:58:39 PST 1998 >Originator: dgaudet@apache.org >Organization: >Release: 1.2,1.3 >Environment: n/a >Description: For various reasons we have defined mechanisms for almost every field in request_rec to be merged or passed down into subrequests, or promoted from a subrequest into the parent request. There is one very notable exception, request_config. This field (not used at all by stock modules) contains arbitrary binary data, and the API does not define any semantics for merging/promoting it. Modules would have an extremely hard time using this field because subrequests are ubiquitous. A similar comment could be made about r->notes. But that is just a table and is easy to define merging semantics without api additions. >How-To-Repeat: >Fix: I tend to lean towards removing request_config. Unless someone can demonstrate a use for it ... in which case we could define some merging functions in the api >Audit-Trail: State-Changed-From-To: open-suspended State-Changed-By: coar State-Changed-When: Sat Dec 6 15:26:21 PST 1997 State-Changed-Why: TBD Class-Changed-From-To: sw-bug-change-request Class-Changed-By: coar Class-Changed-When: Sat Dec 6 15:26:21 PST 1997 Release-Changed-From-To: all through 1.3-1.2,1.3 Release-Changed-By: coar Release-Changed-When: Thu Jan 22 08:43:15 PST 1998 Comment-Added-By: dgaudet Comment-Added-When: Mon Feb 23 03:58:39 PST 1998 Comment-Added: I actually found a use for request_config... a use which relies on it not merging across subrequests. See mod_mmap_static. But in order to use it I still have to double-check that the value I tucked into request_config is still useful later on when I need it again. i.e. request_config is still bogus, and is only useful for saving a binary search in this case. >Unformatted: