tytso@athena.mit.edu (Theodore Y. Ts'o) (06/09/89)
Since snark.UUCP, Eric Raymond's machine, is down, with his permission I'm going to be doing the next release of TMN-news. Part of the problem with TMN-news 7.7 was that it was developed on a small system (A Toshiba Laptop?!?), on a leaf site, and on System V. Working with an almost completely diametrically opposed system --- A MicroVax III, a feed site, and BSD 4.3 --- I have finally gotten the transport layer of TMN-news to a state which I would consider usuable: * It no longer core dumps/crashes * It no longer lower-cases the message-id, thus flooding the network with duplicates if you are a feed another (non-leaf) site. * Unnecessary I/O has been eliminated, making it much, much faster. It is now much faster than News 2.11, and almost as fast as C News. (It does extra things that C News doesn't, such as back chaining of articles.) * It handles control messages correctly, so that MANUALLY and NONEWGROUPS actually work and mean something. * Expire will no longer crash during a rebuilding of the history file if it finds a group that it didn't expect. * Two separate bugs which left garbage entries in the history file and caused expire to complain loadly have been surgically removed. [ Yes, TMNN 7.7 had all of these problems. ] To give you an idea of how much things have improved, I considered the release that Eric called "7.0 Beta" to be pre-Alpha quality code, since it wouldn't even compile with out enormous amounts of hacks and kludges, as I'm sure many people may remember. Well, this new release which I'm planning is what I would consider middle-to-late Beta-quality. I still need to do a few cosmetic things to make it easier to compile and install and to integrate the NNTP changes with the new NNTP 1.5.5 patches that were released recently. However, in the meantime, if anybody else has been playing with TMN-news and has bugfixes to TMNN 7.7, please send them to me. I am particularly interested in patches to make TMNN compile and run on various sundry systems as well as bugfixes, instead of new features. I am intending this release to be a erformance/bugfix release, so we can get something stable out there. The major problem left to be solved is that news readers that come with TMNN still don't quite work on BSD systems; the problem seems to be in the terminal/curses package which was supplied. If some kind BSD soul has worked on it, or has considered working on it, I'd appreciate it a lot. In the past, this has been on the back burner, since rrn and gnus via NNTP has been much easier and used by most people. However, these new readers have a lot more features and have a hypertext-like "conversation following" commands that actually take advantage of the References line that allow you to follow a thread backwards and forwards in coversation order. In addition, once they work, these new readers can also work remotely over NNTP. So this is clearly the next area to be attacked. So in conclusion, a new release of TMNN that actually works is in the works and should be out the door in approximately two weeks. Much blood, sweat, and tears have been spent in making the transport layer as robust as possible. (Read my lips: No New Features!) If anybody else has bugfixes to TMNN 7.7, please send them to me ASAP. I may have fixed that particular bug already, but then again, I might not have. So please send any changes you needed to get TMNN to compile and run on your system, along with a description about what the bug was and how the patch fixes it. Thanks! =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Theodore Ts'o bloom-beacon!mit-athena!tytso 3 Ames St., Cambridge, MA 02139 tytso@athena.mit.edu Everybody's playing the game, but nobody's rules are the same!