Announce: NCSA HTTPd 1.5b2

[Brandon Long <bl...@uiuc.edu>]

Sorry for the extended time between 1.5b1 and 1.5b2, but there were a lot
of little errors, and some other minor problems.

Feel free to pick up the new beta from the same place as beta 1,
http://hoohoo.ncsa.uiuc.edu/beta-1.5/

We are still attempting to keep up with the HyperNews pages, so feel
free to post your problems, gripes, requests, etc. there.

We are not making a patch available for this version since it was 
so big (the patch was over 100k)

Brandon

>From the CHANGES file:
---------------------------------------------------------------------

Fixes for 1.5 Beta 2
--------------------
*) To fix AddType .cgi completely, now has a single virtual path for
   any method to a "standard document" and checks just before sending
   if the method is valid for the document type.
*) General code reorganization, including renaming of scripts to cgi
   and removing all method specific files, and creating an http_send
   for the new general case
*) Moved all non-security compile time options to config.h
*) Support for profiling the code added
*) A/UX, SVR4, SCO3 support updated
*) New type of process title changing subroutine added, and made a compile
   time option
*) Collapsed number of malloc()'s for performance
*) DBM support now properly #define'd
*) Duh, always initialize url/args/filename
*) fixed multihome support to allow more than 2 entries
*) fixed access control
*) log missing user attempt - Gioacchino La Vecchia (gio@di.unipi.it)
*) explicitly allow or disallow certain server configuration directives
   in virtual hosts
*) strip ending / on document root at configuration time
*) stat filesystem for .htaccess files instead of checking failure of fopen()
   for speed improvement
*) change order of mime initialization to get actual location of file first
*) missing LF in redirect message from die()
*) actually log the error message if using an error document for 500 errors
*) close more file descriptors on errors
*) make separate content_type variables for type as sent and type of reply
*) allow Connection header to be made into a CGI variable
*) fix signal handling during CGI 
*) Only shutdown(sd,0) to disallow receives not sends and receives
   This fix was for OSF/1 2.0, but may fix AIX problems of garbling ends
   of images
*) attempt to lessen the number of fflush()'s called during a transaction
*) move local variables that might be clobbered by setjmp/longjmp
*) call tzset on all systems
*) log errno on some errors
*) log which URL is being aborted or timing out
*) Keberos fixes
*) Multiple referer ignore added
*) Probably some I've missed


-- 
 Brandon Long		"I think, therefore I am Confused." - RAW
 HTTPd/SDG/NCSA/UIUC  "Here's a nickel, kid.  Get yourself a better computer."
 blong@uiuc.edu			-"Dilbert" by Scott Adams 6-24-95
 http://www.uiuc.edu/ph/www/blong	Consider myself properly disclaimed.