Return-Path: Delivered-To: apmail-apr-dev-archive@apr.apache.org Received: (qmail 18307 invoked by uid 500); 20 Feb 2003 23:39:08 -0000 Mailing-List: contact dev-help@apr.apache.org; run by ezmlm Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: Delivered-To: mailing list dev@apr.apache.org Received: (qmail 18296 invoked from network); 20 Feb 2003 23:39:07 -0000 Message-ID: <3E556713.7010403@apache.org> Date: Thu, 20 Feb 2003 15:38:59 -0800 From: Ian Holsman User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3b) Gecko/20030211 X-Accept-Language: en-us, en MIME-Version: 1.0 To: trawick@attglobal.net, dev@apr.apache.org Subject: Re: [PATCH] clean up debug patch used with Apache 2 on daedalus References: <3E55193F.3000507@attglobal.net> In-Reply-To: <3E55193F.3000507@attglobal.net> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N +1 from me. now all we need is a directive in httpd to turn this feature on I guess. Jeff Trawick wrote: > Network apps can all too often be tricked by clients into segfaulting or > worse. When this occurs the first time, it can be important to know > what data was received and with what boundaries (i.e., how much was > passed up at a time to the app). > > For quite a while (>1.5 years I think) www.apache.org has run with a > patch to APR which saves a certain amount of data read from the client. > It is then used for debugging in the case of a segfault. Having access > to the actual data and in what segments it was returned to Apache helped > fix some bugs pretty quickly. > > I've cleaned up that patch a bit in hopes of showing how the > functionality could be added to APR. In this new patch, the saving of > data is turned on by the app via a socket option rather than always > enabled as in the www.apache.org patch. > > Is anybody for/against putting something like this in APR? > > It would be nice to enable the app to change the parameters (how many > buffers to save, how much of each buffer to save). I guess those could > be separate options in the future, with the current constants "20" and > "1024" simply the defaults. > > It would also be nice to allow the app to retrieve the list of saved > input buffers, though it is still useful even without that feature.