- From: olga <olga@goliath.eai.com>
- Date: Tue, 26 Jan 1999 09:21:10 -0600 (CST)
- To: Henrik Frystyk Nielsen <frystyk@w3.org>
- Cc: olga <olga@goliath.eai.com>, www-lib@w3.org
Hi, As I understand, on starting the w3c library the .index is read and anchors are created for all the files in cache. When the HT_C_IMS header is created - an anchor for the location is referenced for LM date. In those parts everything goes throug the library - I have not changed anything there... There is what is in the meta-file: olga@nile 58: more AAAa0000a.meta Content-Type: application/x-jt Content-Length: 5844 Accept-Ranges: bytes ETag: "3c000ac-16d4-0 Last-Modified: Thu, 01 Jan 1970 00:00:00 GMT Server: Apache/1.3.1 (Unix) Date: Mon, 25 Jan 1999 22:32:45 GMT --------------------------------------------------- Here is what it is parsed to later on : ----------------------------------------------------------- TTP... in HTTPEvent HTTP_BEGIN: === Connecting to server === HTHost parse Looking up `kingkong' on port 1080 Host info... REUSING CHANNEL 1002c868 Host info... Added Net 100309f0 (request 1002c4c0) to pipe on Host 1002c7b8, 2 requests made, 1 requests in pipe, 0 pending Timer....... Deleted timer 10028c50 HTHost...... No ActivateRequest callback handler registered Channel..... Semaphore increased to 1 for channel 1002c868 Host connect Unlocking Host 1002c7b8 HTTP_NEED_STREAM: === Constructing stream === StreamStack. Constructing stream stack for text/x-http to */* Tee......... Created stream 10048468 with resolver 45c770 HTTP........ Dumping request to `w3chttp.out' HTTP_CONNECTED, HTEvent_WRITE: ===== Flushing request === HTTP........ If-Modified-Since, lm = -3600 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ HTTP........ Generating HTTP/1.x Request Headers HTTP........ Generating General Headers Buffer...... Flushing 1002ae40 .... ------------------------------------------------------------- And here is the corresponding entry in w3chttp.out: GET /projects/Demos/fishing_reel/fishing_reel.jt HTTP/1.1 Accept: */* TE: trailers Authorization: Basic ZWFpOmVhaQ== Host: kingkong:1080 User-Agent: VisNetwork/1.1 libwww/5.1l Connection: TE,Keep-Alive The reasons that I am not using the HTCacheFilter is that I want to read directly from cache (for efficiency) if the file is there and cache verification scheme is currently set to "not verify". This was working in previous version of library I had. Now, when I got newer version (2 weeks ago) it stopped working (I think in part of parsing the date, but I might be wrong of course). Thanks, Olga Antropova. On 25-Jan-99 Henrik Frystyk Nielsen wrote: > > > olga wrote: > >> I am not using the HTCacheFilter and instead add the header HT_C_IMS (If- >> modified-since) to the GET request. Often the header is not added because >> the >> Last-modified field in anchor is parsed to a negative value. Why is that >> happening? > > Unless there is something wrong in the date parser (or the format) then > it is because it doesn't have a LM date and hence it returns -1. What > the persistent cache allows you to do is to use etags instead and to > maintain these etags persistently so that you can start the app at a > later time and still be able to use conditional GET requests. > >> Also even if Last-modified is parsed to positive value, and >> If-modified-since >> is added to headers the file is still loaded from server and 200 is >> returned. > > If the server sends back a 200 then it is because it thinks that the > document has changed. > > Is there any reason why you can't use the normal libwww persistent > cache? Amaya, Web Commander, and the webbot are all using it. > > Henrik > > -- > Henrik Frystyk Nielsen, > World Wide Web Consortium > http://www.w3.org/People/Frystyk ---------------------------------- E-Mail: olga <olga@eai.com> Date: 25-Jan-99 Time: 16:33:52 This message was sent by XFMail ----------------------------------
Received on Tuesday, 26 January 1999 10:17:28 UTC